Hugging Face's logo

trueorfalse441
/
whisper-small-ko-custom-1000h

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
whisper
Model card Files Metrics Community
trueorfalse441 commited on
Commit
8004826
·
1 Parent(s): d117b87

Upload 54 files

Browse files
Files changed (47) hide show
  1. checkpoint-1000/config.json +51 -0
  2. checkpoint-1000/generation_config.json +221 -0
  3. checkpoint-1000/optimizer.pt +3 -0
  4. checkpoint-1000/preprocessor_config.json +14 -0
  5. checkpoint-1000/pytorch_model.bin +3 -0
  6. checkpoint-1000/rng_state.pth +3 -0
  7. checkpoint-1000/scheduler.pt +3 -0
  8. checkpoint-1000/trainer_state.json +265 -0
  9. checkpoint-1000/training_args.bin +3 -0
  10. checkpoint-2000/config.json +51 -0
  11. checkpoint-2000/generation_config.json +221 -0
  12. checkpoint-2000/optimizer.pt +3 -0
  13. checkpoint-2000/preprocessor_config.json +14 -0
  14. checkpoint-2000/pytorch_model.bin +3 -0
  15. checkpoint-2000/rng_state.pth +3 -0
  16. checkpoint-2000/scheduler.pt +3 -0
  17. checkpoint-2000/trainer_state.json +514 -0
  18. checkpoint-2000/training_args.bin +3 -0
  19. checkpoint-3000/config.json +51 -0
  20. checkpoint-3000/generation_config.json +221 -0
  21. checkpoint-3000/optimizer.pt +3 -0
  22. checkpoint-3000/preprocessor_config.json +14 -0
  23. checkpoint-3000/pytorch_model.bin +3 -0
  24. checkpoint-3000/rng_state.pth +3 -0
  25. checkpoint-3000/scheduler.pt +3 -0
  26. checkpoint-3000/trainer_state.json +763 -0
  27. checkpoint-3000/training_args.bin +3 -0
  28. checkpoint-4000/config.json +51 -0
  29. checkpoint-4000/generation_config.json +221 -0
  30. checkpoint-4000/optimizer.pt +3 -0
  31. checkpoint-4000/preprocessor_config.json +14 -0
  32. checkpoint-4000/pytorch_model.bin +3 -0
  33. checkpoint-4000/rng_state.pth +3 -0
  34. checkpoint-4000/scheduler.pt +3 -0
  35. checkpoint-4000/trainer_state.json +1012 -0
  36. checkpoint-4000/training_args.bin +3 -0
  37. checkpoint-5000/config.json +51 -0
  38. checkpoint-5000/generation_config.json +221 -0
  39. checkpoint-5000/optimizer.pt +3 -0
  40. checkpoint-5000/preprocessor_config.json +14 -0
  41. checkpoint-5000/pytorch_model.bin +3 -0
  42. checkpoint-5000/rng_state.pth +3 -0
  43. checkpoint-5000/scheduler.pt +3 -0
  44. checkpoint-5000/trainer_state.json +1261 -0
  45. checkpoint-5000/training_args.bin +3 -0
  46. generation_config.json +221 -0
  47. runs/Apr27_17-49-52_AI-2nd/events.out.tfevents.1682585399.AI-2nd.11600.0 +2 -2
checkpoint-1000/config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "model_type": "whisper",
41
+ "num_hidden_layers": 12,
42
+ "num_mel_bins": 80,
43
+ "pad_token_id": 50257,
44
+ "scale_embedding": false,
45
+ "suppress_tokens": [],
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.28.1",
48
+ "use_cache": true,
49
+ "use_weighted_layer_sum": false,
50
+ "vocab_size": 51865
51
+ }
checkpoint-1000/generation_config.json ADDED
@@ -0,0 +1,221 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "begin_suppress_tokens": [
3
+ 220,
4
+ 50257
5
+ ],
6
+ "bos_token_id": 50257,
7
+ "decoder_start_token_id": 50258,
8
+ "eos_token_id": 50257,
9
+ "forced_decoder_ids": [
10
+ [
11
+ 1,
12
+ null
13
+ ],
14
+ [
15
+ 2,
16
+ 50359
17
+ ]
18
+ ],
19
+ "is_multilingual": true,
20
+ "lang_to_id": {
21
+ "<|af|>": 50327,
22
+ "<|am|>": 50334,
23
+ "<|ar|>": 50272,
24
+ "<|as|>": 50350,
25
+ "<|az|>": 50304,
26
+ "<|ba|>": 50355,
27
+ "<|be|>": 50330,
28
+ "<|bg|>": 50292,
29
+ "<|bn|>": 50302,
30
+ "<|bo|>": 50347,
31
+ "<|br|>": 50309,
32
+ "<|bs|>": 50315,
33
+ "<|ca|>": 50270,
34
+ "<|cs|>": 50283,
35
+ "<|cy|>": 50297,
36
+ "<|da|>": 50285,
37
+ "<|de|>": 50261,
38
+ "<|el|>": 50281,
39
+ "<|en|>": 50259,
40
+ "<|es|>": 50262,
41
+ "<|et|>": 50307,
42
+ "<|eu|>": 50310,
43
+ "<|fa|>": 50300,
44
+ "<|fi|>": 50277,
45
+ "<|fo|>": 50338,
46
+ "<|fr|>": 50265,
47
+ "<|gl|>": 50319,
48
+ "<|gu|>": 50333,
49
+ "<|haw|>": 50352,
50
+ "<|ha|>": 50354,
51
+ "<|he|>": 50279,
52
+ "<|hi|>": 50276,
53
+ "<|hr|>": 50291,
54
+ "<|ht|>": 50339,
55
+ "<|hu|>": 50286,
56
+ "<|hy|>": 50312,
57
+ "<|id|>": 50275,
58
+ "<|is|>": 50311,
59
+ "<|it|>": 50274,
60
+ "<|ja|>": 50266,
61
+ "<|jw|>": 50356,
62
+ "<|ka|>": 50329,
63
+ "<|kk|>": 50316,
64
+ "<|km|>": 50323,
65
+ "<|kn|>": 50306,
66
+ "<|ko|>": 50264,
67
+ "<|la|>": 50294,
68
+ "<|lb|>": 50345,
69
+ "<|ln|>": 50353,
70
+ "<|lo|>": 50336,
71
+ "<|lt|>": 50293,
72
+ "<|lv|>": 50301,
73
+ "<|mg|>": 50349,
74
+ "<|mi|>": 50295,
75
+ "<|mk|>": 50308,
76
+ "<|ml|>": 50296,
77
+ "<|mn|>": 50314,
78
+ "<|mr|>": 50320,
79
+ "<|ms|>": 50282,
80
+ "<|mt|>": 50343,
81
+ "<|my|>": 50346,
82
+ "<|ne|>": 50313,
83
+ "<|nl|>": 50271,
84
+ "<|nn|>": 50342,
85
+ "<|no|>": 50288,
86
+ "<|oc|>": 50328,
87
+ "<|pa|>": 50321,
88
+ "<|pl|>": 50269,
89
+ "<|ps|>": 50340,
90
+ "<|pt|>": 50267,
91
+ "<|ro|>": 50284,
92
+ "<|ru|>": 50263,
93
+ "<|sa|>": 50344,
94
+ "<|sd|>": 50332,
95
+ "<|si|>": 50322,
96
+ "<|sk|>": 50298,
97
+ "<|sl|>": 50305,
98
+ "<|sn|>": 50324,
99
+ "<|so|>": 50326,
100
+ "<|sq|>": 50317,
101
+ "<|sr|>": 50303,
102
+ "<|su|>": 50357,
103
+ "<|sv|>": 50273,
104
+ "<|sw|>": 50318,
105
+ "<|ta|>": 50287,
106
+ "<|te|>": 50299,
107
+ "<|tg|>": 50331,
108
+ "<|th|>": 50289,
109
+ "<|tk|>": 50341,
110
+ "<|tl|>": 50348,
111
+ "<|tr|>": 50268,
112
+ "<|tt|>": 50351,
113
+ "<|uk|>": 50280,
114
+ "<|ur|>": 50290,
115
+ "<|uz|>": 50337,
116
+ "<|vi|>": 50278,
117
+ "<|yi|>": 50335,
118
+ "<|yo|>": 50325,
119
+ "<|zh|>": 50260
120
+ },
121
+ "max_initial_timestamp_index": 1,
122
+ "max_length": 448,
123
+ "no_timestamps_token_id": 50363,
124
+ "pad_token_id": 50257,
125
+ "return_timestamps": false,
126
+ "suppress_tokens": [
127
+ 1,
128
+ 2,
129
+ 7,
130
+ 8,
131
+ 9,
132
+ 10,
133
+ 14,
134
+ 25,
135
+ 26,
136
+ 27,
137
+ 28,
138
+ 29,
139
+ 31,
140
+ 58,
141
+ 59,
142
+ 60,
143
+ 61,
144
+ 62,
145
+ 63,
146
+ 90,
147
+ 91,
148
+ 92,
149
+ 93,
150
+ 359,
151
+ 503,
152
+ 522,
153
+ 542,
154
+ 873,
155
+ 893,
156
+ 902,
157
+ 918,
158
+ 922,
159
+ 931,
160
+ 1350,
161
+ 1853,
162
+ 1982,
163
+ 2460,
164
+ 2627,
165
+ 3246,
166
+ 3253,
167
+ 3268,
168
+ 3536,
169
+ 3846,
170
+ 3961,
171
+ 4183,
172
+ 4667,
173
+ 6585,
174
+ 6647,
175
+ 7273,
176
+ 9061,
177
+ 9383,
178
+ 10428,
179
+ 10929,
180
+ 11938,
181
+ 12033,
182
+ 12331,
183
+ 12562,
184
+ 13793,
185
+ 14157,
186
+ 14635,
187
+ 15265,
188
+ 15618,
189
+ 16553,
190
+ 16604,
191
+ 18362,
192
+ 18956,
193
+ 20075,
194
+ 21675,
195
+ 22520,
196
+ 26130,
197
+ 26161,
198
+ 26435,
199
+ 28279,
200
+ 29464,
201
+ 31650,
202
+ 32302,
203
+ 32470,
204
+ 36865,
205
+ 42863,
206
+ 47425,
207
+ 49870,
208
+ 50254,
209
+ 50258,
210
+ 50358,
211
+ 50359,
212
+ 50360,
213
+ 50361,
214
+ 50362
215
+ ],
216
+ "task_to_id": {
217
+ "transcribe": 50359,
218
+ "translate": 50358
219
+ },
220
+ "transformers_version": "4.28.1"
221
+ }
checkpoint-1000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d5c6c82bb201da39727c5849a49084d6bc11281447e405c9881c7e07f93a8ad2
3
+ size 1934161093
checkpoint-1000/preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 80,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
checkpoint-1000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:50088474fe93aa060ec5be3057aa01ac118700951f572f9491a70aac2d5bb759
3
+ size 967102601
checkpoint-1000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95bdaa9937d0a79a54ba454fc6e9f5c986a7b76b8207f126f2c3c9f9b852a1e9
3
+ size 14575
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a8e8891efa973f5e42eb1308db86123e1a5680f28385d58e29344b5b45497064
3
+ size 627
checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,265 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 33.16084938138103,
3
+ "best_model_checkpoint": "./whisper-small-ko-custom-1000h\\checkpoint-1000",
4
+ "epoch": 0.00669290284582229,
5
+ "global_step": 1000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 2.5e-06,
13
+ "loss": 3.5773,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 5e-06,
19
+ "loss": 1.4784,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.0,
24
+ "learning_rate": 7.500000000000001e-06,
25
+ "loss": 0.753,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.0,
30
+ "learning_rate": 1e-05,
31
+ "loss": 0.5502,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 0.0,
36
+ "learning_rate": 9.948979591836737e-06,
37
+ "loss": 0.3562,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 0.0,
42
+ "learning_rate": 9.89795918367347e-06,
43
+ "loss": 0.2842,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 0.0,
48
+ "learning_rate": 9.846938775510205e-06,
49
+ "loss": 0.2484,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 0.0,
54
+ "learning_rate": 9.795918367346939e-06,
55
+ "loss": 0.2421,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 0.0,
60
+ "learning_rate": 9.744897959183674e-06,
61
+ "loss": 0.2403,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 0.0,
66
+ "learning_rate": 9.693877551020408e-06,
67
+ "loss": 0.2222,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 0.0,
72
+ "learning_rate": 9.642857142857144e-06,
73
+ "loss": 0.202,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 0.0,
78
+ "learning_rate": 9.591836734693878e-06,
79
+ "loss": 0.2197,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 0.0,
84
+ "learning_rate": 9.540816326530612e-06,
85
+ "loss": 0.2045,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 0.0,
90
+ "learning_rate": 9.489795918367348e-06,
91
+ "loss": 0.2275,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 0.0,
96
+ "learning_rate": 9.438775510204082e-06,
97
+ "loss": 0.2023,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 0.0,
102
+ "learning_rate": 9.387755102040818e-06,
103
+ "loss": 0.2212,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 0.0,
108
+ "learning_rate": 9.336734693877552e-06,
109
+ "loss": 0.211,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 0.0,
114
+ "learning_rate": 9.285714285714288e-06,
115
+ "loss": 0.2041,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 0.0,
120
+ "learning_rate": 9.234693877551022e-06,
121
+ "loss": 0.2215,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 0.0,
126
+ "learning_rate": 9.183673469387756e-06,
127
+ "loss": 0.2012,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 0.0,
132
+ "learning_rate": 9.13265306122449e-06,
133
+ "loss": 0.1932,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 0.0,
138
+ "learning_rate": 9.081632653061225e-06,
139
+ "loss": 0.2087,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 0.0,
144
+ "learning_rate": 9.03061224489796e-06,
145
+ "loss": 0.1908,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 0.0,
150
+ "learning_rate": 8.979591836734695e-06,
151
+ "loss": 0.2019,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 0.0,
156
+ "learning_rate": 8.92857142857143e-06,
157
+ "loss": 0.2097,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 0.0,
162
+ "learning_rate": 8.877551020408163e-06,
163
+ "loss": 0.2105,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 0.0,
168
+ "learning_rate": 8.826530612244899e-06,
169
+ "loss": 0.1846,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 0.0,
174
+ "learning_rate": 8.775510204081633e-06,
175
+ "loss": 0.1748,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 0.0,
180
+ "learning_rate": 8.724489795918369e-06,
181
+ "loss": 0.1761,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 0.01,
186
+ "learning_rate": 8.673469387755103e-06,
187
+ "loss": 0.1982,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 0.01,
192
+ "learning_rate": 8.622448979591837e-06,
193
+ "loss": 0.1926,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 0.01,
198
+ "learning_rate": 8.571428571428571e-06,
199
+ "loss": 0.1732,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 0.01,
204
+ "learning_rate": 8.520408163265307e-06,
205
+ "loss": 0.1841,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 0.01,
210
+ "learning_rate": 8.469387755102042e-06,
211
+ "loss": 0.1699,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 0.01,
216
+ "learning_rate": 8.418367346938776e-06,
217
+ "loss": 0.2131,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 0.01,
222
+ "learning_rate": 8.36734693877551e-06,
223
+ "loss": 0.1828,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 0.01,
228
+ "learning_rate": 8.316326530612246e-06,
229
+ "loss": 0.1796,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 0.01,
234
+ "learning_rate": 8.26530612244898e-06,
235
+ "loss": 0.1909,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 0.01,
240
+ "learning_rate": 8.214285714285714e-06,
241
+ "loss": 0.2066,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 0.01,
246
+ "learning_rate": 8.16326530612245e-06,
247
+ "loss": 0.1913,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 0.01,
252
+ "eval_loss": 0.17891158163547516,
253
+ "eval_runtime": 14046.7097,
254
+ "eval_samples_per_second": 3.473,
255
+ "eval_steps_per_second": 0.434,
256
+ "eval_wer": 33.16084938138103,
257
+ "step": 1000
258
+ }
259
+ ],
260
+ "max_steps": 5000,
261
+ "num_train_epochs": 1,
262
+ "total_flos": 4.61736640512e+18,
263
+ "trial_name": null,
264
+ "trial_params": null
265
+ }
checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe99781f928141fde56544c833b4f8a8a1100a885088c1e45942567c2ed1ebfc
3
+ size 3771
checkpoint-2000/config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "model_type": "whisper",
41
+ "num_hidden_layers": 12,
42
+ "num_mel_bins": 80,
43
+ "pad_token_id": 50257,
44
+ "scale_embedding": false,
45
+ "suppress_tokens": [],
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.28.1",
48
+ "use_cache": true,
49
+ "use_weighted_layer_sum": false,
50
+ "vocab_size": 51865
51
+ }
checkpoint-2000/generation_config.json ADDED
@@ -0,0 +1,221 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "begin_suppress_tokens": [
3
+ 220,
4
+ 50257
5
+ ],
6
+ "bos_token_id": 50257,
7
+ "decoder_start_token_id": 50258,
8
+ "eos_token_id": 50257,
9
+ "forced_decoder_ids": [
10
+ [
11
+ 1,
12
+ null
13
+ ],
14
+ [
15
+ 2,
16
+ 50359
17
+ ]
18
+ ],
19
+ "is_multilingual": true,
20
+ "lang_to_id": {
21
+ "<|af|>": 50327,
22
+ "<|am|>": 50334,
23
+ "<|ar|>": 50272,
24
+ "<|as|>": 50350,
25
+ "<|az|>": 50304,
26
+ "<|ba|>": 50355,
27
+ "<|be|>": 50330,
28
+ "<|bg|>": 50292,
29
+ "<|bn|>": 50302,
30
+ "<|bo|>": 50347,
31
+ "<|br|>": 50309,
32
+ "<|bs|>": 50315,
33
+ "<|ca|>": 50270,
34
+ "<|cs|>": 50283,
35
+ "<|cy|>": 50297,
36
+ "<|da|>": 50285,
37
+ "<|de|>": 50261,
38
+ "<|el|>": 50281,
39
+ "<|en|>": 50259,
40
+ "<|es|>": 50262,
41
+ "<|et|>": 50307,
42
+ "<|eu|>": 50310,
43
+ "<|fa|>": 50300,
44
+ "<|fi|>": 50277,
45
+ "<|fo|>": 50338,
46
+ "<|fr|>": 50265,
47
+ "<|gl|>": 50319,
48
+ "<|gu|>": 50333,
49
+ "<|haw|>": 50352,
50
+ "<|ha|>": 50354,
51
+ "<|he|>": 50279,
52
+ "<|hi|>": 50276,
53
+ "<|hr|>": 50291,
54
+ "<|ht|>": 50339,
55
+ "<|hu|>": 50286,
56
+ "<|hy|>": 50312,
57
+ "<|id|>": 50275,
58
+ "<|is|>": 50311,
59
+ "<|it|>": 50274,
60
+ "<|ja|>": 50266,
61
+ "<|jw|>": 50356,
62
+ "<|ka|>": 50329,
63
+ "<|kk|>": 50316,
64
+ "<|km|>": 50323,
65
+ "<|kn|>": 50306,
66
+ "<|ko|>": 50264,
67
+ "<|la|>": 50294,
68
+ "<|lb|>": 50345,
69
+ "<|ln|>": 50353,
70
+ "<|lo|>": 50336,
71
+ "<|lt|>": 50293,
72
+ "<|lv|>": 50301,
73
+ "<|mg|>": 50349,
74
+ "<|mi|>": 50295,
75
+ "<|mk|>": 50308,
76
+ "<|ml|>": 50296,
77
+ "<|mn|>": 50314,
78
+ "<|mr|>": 50320,
79
+ "<|ms|>": 50282,
80
+ "<|mt|>": 50343,
81
+ "<|my|>": 50346,
82
+ "<|ne|>": 50313,
83
+ "<|nl|>": 50271,
84
+ "<|nn|>": 50342,
85
+ "<|no|>": 50288,
86
+ "<|oc|>": 50328,
87
+ "<|pa|>": 50321,
88
+ "<|pl|>": 50269,
89
+ "<|ps|>": 50340,
90
+ "<|pt|>": 50267,
91
+ "<|ro|>": 50284,
92
+ "<|ru|>": 50263,
93
+ "<|sa|>": 50344,
94
+ "<|sd|>": 50332,
95
+ "<|si|>": 50322,
96
+ "<|sk|>": 50298,
97
+ "<|sl|>": 50305,
98
+ "<|sn|>": 50324,
99
+ "<|so|>": 50326,
100
+ "<|sq|>": 50317,
101
+ "<|sr|>": 50303,
102
+ "<|su|>": 50357,
103
+ "<|sv|>": 50273,
104
+ "<|sw|>": 50318,
105
+ "<|ta|>": 50287,
106
+ "<|te|>": 50299,
107
+ "<|tg|>": 50331,
108
+ "<|th|>": 50289,
109
+ "<|tk|>": 50341,
110
+ "<|tl|>": 50348,
111
+ "<|tr|>": 50268,
112
+ "<|tt|>": 50351,
113
+ "<|uk|>": 50280,
114
+ "<|ur|>": 50290,
115
+ "<|uz|>": 50337,
116
+ "<|vi|>": 50278,
117
+ "<|yi|>": 50335,
118
+ "<|yo|>": 50325,
119
+ "<|zh|>": 50260
120
+ },
121
+ "max_initial_timestamp_index": 1,
122
+ "max_length": 448,
123
+ "no_timestamps_token_id": 50363,
124
+ "pad_token_id": 50257,
125
+ "return_timestamps": false,
126
+ "suppress_tokens": [
127
+ 1,
128
+ 2,
129
+ 7,
130
+ 8,
131
+ 9,
132
+ 10,
133
+ 14,
134
+ 25,
135
+ 26,
136
+ 27,
137
+ 28,
138
+ 29,
139
+ 31,
140
+ 58,
141
+ 59,
142
+ 60,
143
+ 61,
144
+ 62,
145
+ 63,
146
+ 90,
147
+ 91,
148
+ 92,
149
+ 93,
150
+ 359,
151
+ 503,
152
+ 522,
153
+ 542,
154
+ 873,
155
+ 893,
156
+ 902,
157
+ 918,
158
+ 922,
159
+ 931,
160
+ 1350,
161
+ 1853,
162
+ 1982,
163
+ 2460,
164
+ 2627,
165
+ 3246,
166
+ 3253,
167
+ 3268,
168
+ 3536,
169
+ 3846,
170
+ 3961,
171
+ 4183,
172
+ 4667,
173
+ 6585,
174
+ 6647,
175
+ 7273,
176
+ 9061,
177
+ 9383,
178
+ 10428,
179
+ 10929,
180
+ 11938,
181
+ 12033,
182
+ 12331,
183
+ 12562,
184
+ 13793,
185
+ 14157,
186
+ 14635,
187
+ 15265,
188
+ 15618,
189
+ 16553,
190
+ 16604,
191
+ 18362,
192
+ 18956,
193
+ 20075,
194
+ 21675,
195
+ 22520,
196
+ 26130,
197
+ 26161,
198
+ 26435,
199
+ 28279,
200
+ 29464,
201
+ 31650,
202
+ 32302,
203
+ 32470,
204
+ 36865,
205
+ 42863,
206
+ 47425,
207
+ 49870,
208
+ 50254,
209
+ 50258,
210
+ 50358,
211
+ 50359,
212
+ 50360,
213
+ 50361,
214
+ 50362
215
+ ],
216
+ "task_to_id": {
217
+ "transcribe": 50359,
218
+ "translate": 50358
219
+ },
220
+ "transformers_version": "4.28.1"
221
+ }
checkpoint-2000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52d5c5a5b2060cde6145075f52e815f048fedbba918d035ff06e088daf96e6fa
3
+ size 1934161093
checkpoint-2000/preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 80,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
checkpoint-2000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7e54054daccca9eac1506c6f179bc7ceb72124f8c8d0af4f55bb31bcbd7be916
3
+ size 967102601
checkpoint-2000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d65deae752de99fd45be8260b47aaf49d43a44a717586ab016cf3d0e4aa3300c
3
+ size 14575
checkpoint-2000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1af1facaf3f522d67fba7fb65ab313719109543273fa488df2f4201d4a80aa8e
3
+ size 627
checkpoint-2000/trainer_state.json ADDED
@@ -0,0 +1,514 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 33.16084938138103,
3
+ "best_model_checkpoint": "./whisper-small-ko-custom-1000h\\checkpoint-1000",
4
+ "epoch": 0.01338580569164458,
5
+ "global_step": 2000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 2.5e-06,
13
+ "loss": 3.5773,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 5e-06,
19
+ "loss": 1.4784,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.0,
24
+ "learning_rate": 7.500000000000001e-06,
25
+ "loss": 0.753,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.0,
30
+ "learning_rate": 1e-05,
31
+ "loss": 0.5502,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 0.0,
36
+ "learning_rate": 9.948979591836737e-06,
37
+ "loss": 0.3562,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 0.0,
42
+ "learning_rate": 9.89795918367347e-06,
43
+ "loss": 0.2842,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 0.0,
48
+ "learning_rate": 9.846938775510205e-06,
49
+ "loss": 0.2484,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 0.0,
54
+ "learning_rate": 9.795918367346939e-06,
55
+ "loss": 0.2421,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 0.0,
60
+ "learning_rate": 9.744897959183674e-06,
61
+ "loss": 0.2403,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 0.0,
66
+ "learning_rate": 9.693877551020408e-06,
67
+ "loss": 0.2222,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 0.0,
72
+ "learning_rate": 9.642857142857144e-06,
73
+ "loss": 0.202,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 0.0,
78
+ "learning_rate": 9.591836734693878e-06,
79
+ "loss": 0.2197,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 0.0,
84
+ "learning_rate": 9.540816326530612e-06,
85
+ "loss": 0.2045,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 0.0,
90
+ "learning_rate": 9.489795918367348e-06,
91
+ "loss": 0.2275,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 0.0,
96
+ "learning_rate": 9.438775510204082e-06,
97
+ "loss": 0.2023,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 0.0,
102
+ "learning_rate": 9.387755102040818e-06,
103
+ "loss": 0.2212,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 0.0,
108
+ "learning_rate": 9.336734693877552e-06,
109
+ "loss": 0.211,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 0.0,
114
+ "learning_rate": 9.285714285714288e-06,
115
+ "loss": 0.2041,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 0.0,
120
+ "learning_rate": 9.234693877551022e-06,
121
+ "loss": 0.2215,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 0.0,
126
+ "learning_rate": 9.183673469387756e-06,
127
+ "loss": 0.2012,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 0.0,
132
+ "learning_rate": 9.13265306122449e-06,
133
+ "loss": 0.1932,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 0.0,
138
+ "learning_rate": 9.081632653061225e-06,
139
+ "loss": 0.2087,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 0.0,
144
+ "learning_rate": 9.03061224489796e-06,
145
+ "loss": 0.1908,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 0.0,
150
+ "learning_rate": 8.979591836734695e-06,
151
+ "loss": 0.2019,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 0.0,
156
+ "learning_rate": 8.92857142857143e-06,
157
+ "loss": 0.2097,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 0.0,
162
+ "learning_rate": 8.877551020408163e-06,
163
+ "loss": 0.2105,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 0.0,
168
+ "learning_rate": 8.826530612244899e-06,
169
+ "loss": 0.1846,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 0.0,
174
+ "learning_rate": 8.775510204081633e-06,
175
+ "loss": 0.1748,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 0.0,
180
+ "learning_rate": 8.724489795918369e-06,
181
+ "loss": 0.1761,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 0.01,
186
+ "learning_rate": 8.673469387755103e-06,
187
+ "loss": 0.1982,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 0.01,
192
+ "learning_rate": 8.622448979591837e-06,
193
+ "loss": 0.1926,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 0.01,
198
+ "learning_rate": 8.571428571428571e-06,
199
+ "loss": 0.1732,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 0.01,
204
+ "learning_rate": 8.520408163265307e-06,
205
+ "loss": 0.1841,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 0.01,
210
+ "learning_rate": 8.469387755102042e-06,
211
+ "loss": 0.1699,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 0.01,
216
+ "learning_rate": 8.418367346938776e-06,
217
+ "loss": 0.2131,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 0.01,
222
+ "learning_rate": 8.36734693877551e-06,
223
+ "loss": 0.1828,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 0.01,
228
+ "learning_rate": 8.316326530612246e-06,
229
+ "loss": 0.1796,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 0.01,
234
+ "learning_rate": 8.26530612244898e-06,
235
+ "loss": 0.1909,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 0.01,
240
+ "learning_rate": 8.214285714285714e-06,
241
+ "loss": 0.2066,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 0.01,
246
+ "learning_rate": 8.16326530612245e-06,
247
+ "loss": 0.1913,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 0.01,
252
+ "eval_loss": 0.17891158163547516,
253
+ "eval_runtime": 14046.7097,
254
+ "eval_samples_per_second": 3.473,
255
+ "eval_steps_per_second": 0.434,
256
+ "eval_wer": 33.16084938138103,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 0.01,
261
+ "learning_rate": 8.112244897959184e-06,
262
+ "loss": 0.1426,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 0.01,
267
+ "learning_rate": 8.06122448979592e-06,
268
+ "loss": 0.1976,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 0.01,
273
+ "learning_rate": 8.010204081632654e-06,
274
+ "loss": 0.1677,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 0.01,
279
+ "learning_rate": 7.959183673469388e-06,
280
+ "loss": 0.1633,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 0.01,
285
+ "learning_rate": 7.908163265306124e-06,
286
+ "loss": 0.1994,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 0.01,
291
+ "learning_rate": 7.857142857142858e-06,
292
+ "loss": 0.162,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 0.01,
297
+ "learning_rate": 7.806122448979593e-06,
298
+ "loss": 0.1779,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 0.01,
303
+ "learning_rate": 7.755102040816327e-06,
304
+ "loss": 0.1829,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 0.01,
309
+ "learning_rate": 7.704081632653061e-06,
310
+ "loss": 0.1933,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 0.01,
315
+ "learning_rate": 7.653061224489796e-06,
316
+ "loss": 0.1606,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 0.01,
321
+ "learning_rate": 7.602040816326531e-06,
322
+ "loss": 0.1785,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 0.01,
327
+ "learning_rate": 7.551020408163265e-06,
328
+ "loss": 0.1802,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 0.01,
333
+ "learning_rate": 7.500000000000001e-06,
334
+ "loss": 0.17,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 0.01,
339
+ "learning_rate": 7.448979591836736e-06,
340
+ "loss": 0.1679,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 0.01,
345
+ "learning_rate": 7.39795918367347e-06,
346
+ "loss": 0.157,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 0.01,
351
+ "learning_rate": 7.346938775510205e-06,
352
+ "loss": 0.161,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 0.01,
357
+ "learning_rate": 7.295918367346939e-06,
358
+ "loss": 0.1494,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 0.01,
363
+ "learning_rate": 7.244897959183675e-06,
364
+ "loss": 0.1799,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 0.01,
369
+ "learning_rate": 7.193877551020409e-06,
370
+ "loss": 0.1668,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 0.01,
375
+ "learning_rate": 7.1428571428571436e-06,
376
+ "loss": 0.1798,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 0.01,
381
+ "learning_rate": 7.091836734693878e-06,
382
+ "loss": 0.1476,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 0.01,
387
+ "learning_rate": 7.0408163265306125e-06,
388
+ "loss": 0.1779,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 0.01,
393
+ "learning_rate": 6.989795918367348e-06,
394
+ "loss": 0.1795,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 0.01,
399
+ "learning_rate": 6.938775510204082e-06,
400
+ "loss": 0.1642,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 0.01,
405
+ "learning_rate": 6.887755102040817e-06,
406
+ "loss": 0.1239,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 0.01,
411
+ "learning_rate": 6.836734693877551e-06,
412
+ "loss": 0.2013,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 0.01,
417
+ "learning_rate": 6.785714285714287e-06,
418
+ "loss": 0.1646,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 0.01,
423
+ "learning_rate": 6.734693877551021e-06,
424
+ "loss": 0.1339,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 0.01,
429
+ "learning_rate": 6.683673469387756e-06,
430
+ "loss": 0.1814,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 0.01,
435
+ "learning_rate": 6.63265306122449e-06,
436
+ "loss": 0.1302,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 0.01,
441
+ "learning_rate": 6.581632653061225e-06,
442
+ "loss": 0.1324,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 0.01,
447
+ "learning_rate": 6.530612244897959e-06,
448
+ "loss": 0.1384,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 0.01,
453
+ "learning_rate": 6.4795918367346946e-06,
454
+ "loss": 0.1628,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 0.01,
459
+ "learning_rate": 6.4285714285714295e-06,
460
+ "loss": 0.1712,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 0.01,
465
+ "learning_rate": 6.3775510204081635e-06,
466
+ "loss": 0.153,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 0.01,
471
+ "learning_rate": 6.326530612244899e-06,
472
+ "loss": 0.1588,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 0.01,
477
+ "learning_rate": 6.275510204081633e-06,
478
+ "loss": 0.1579,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 0.01,
483
+ "learning_rate": 6.224489795918368e-06,
484
+ "loss": 0.2014,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 0.01,
489
+ "learning_rate": 6.173469387755102e-06,
490
+ "loss": 0.153,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 0.01,
495
+ "learning_rate": 6.122448979591837e-06,
496
+ "loss": 0.141,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 0.01,
501
+ "eval_loss": 0.153132826089859,
502
+ "eval_runtime": 14050.6821,
503
+ "eval_samples_per_second": 3.472,
504
+ "eval_steps_per_second": 0.434,
505
+ "eval_wer": 33.639439546838894,
506
+ "step": 2000
507
+ }
508
+ ],
509
+ "max_steps": 5000,
510
+ "num_train_epochs": 1,
511
+ "total_flos": 9.23473281024e+18,
512
+ "trial_name": null,
513
+ "trial_params": null
514
+ }
checkpoint-2000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe99781f928141fde56544c833b4f8a8a1100a885088c1e45942567c2ed1ebfc
3
+ size 3771
checkpoint-3000/config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "model_type": "whisper",
41
+ "num_hidden_layers": 12,
42
+ "num_mel_bins": 80,
43
+ "pad_token_id": 50257,
44
+ "scale_embedding": false,
45
+ "suppress_tokens": [],
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.28.1",
48
+ "use_cache": true,
49
+ "use_weighted_layer_sum": false,
50
+ "vocab_size": 51865
51
+ }
checkpoint-3000/generation_config.json ADDED
@@ -0,0 +1,221 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "begin_suppress_tokens": [
3
+ 220,
4
+ 50257
5
+ ],
6
+ "bos_token_id": 50257,
7
+ "decoder_start_token_id": 50258,
8
+ "eos_token_id": 50257,
9
+ "forced_decoder_ids": [
10
+ [
11
+ 1,
12
+ null
13
+ ],
14
+ [
15
+ 2,
16
+ 50359
17
+ ]
18
+ ],
19
+ "is_multilingual": true,
20
+ "lang_to_id": {
21
+ "<|af|>": 50327,
22
+ "<|am|>": 50334,
23
+ "<|ar|>": 50272,
24
+ "<|as|>": 50350,
25
+ "<|az|>": 50304,
26
+ "<|ba|>": 50355,
27
+ "<|be|>": 50330,
28
+ "<|bg|>": 50292,
29
+ "<|bn|>": 50302,
30
+ "<|bo|>": 50347,
31
+ "<|br|>": 50309,
32
+ "<|bs|>": 50315,
33
+ "<|ca|>": 50270,
34
+ "<|cs|>": 50283,
35
+ "<|cy|>": 50297,
36
+ "<|da|>": 50285,
37
+ "<|de|>": 50261,
38
+ "<|el|>": 50281,
39
+ "<|en|>": 50259,
40
+ "<|es|>": 50262,
41
+ "<|et|>": 50307,
42
+ "<|eu|>": 50310,
43
+ "<|fa|>": 50300,
44
+ "<|fi|>": 50277,
45
+ "<|fo|>": 50338,
46
+ "<|fr|>": 50265,
47
+ "<|gl|>": 50319,
48
+ "<|gu|>": 50333,
49
+ "<|haw|>": 50352,
50
+ "<|ha|>": 50354,
51
+ "<|he|>": 50279,
52
+ "<|hi|>": 50276,
53
+ "<|hr|>": 50291,
54
+ "<|ht|>": 50339,
55
+ "<|hu|>": 50286,
56
+ "<|hy|>": 50312,
57
+ "<|id|>": 50275,
58
+ "<|is|>": 50311,
59
+ "<|it|>": 50274,
60
+ "<|ja|>": 50266,
61
+ "<|jw|>": 50356,
62
+ "<|ka|>": 50329,
63
+ "<|kk|>": 50316,
64
+ "<|km|>": 50323,
65
+ "<|kn|>": 50306,
66
+ "<|ko|>": 50264,
67
+ "<|la|>": 50294,
68
+ "<|lb|>": 50345,
69
+ "<|ln|>": 50353,
70
+ "<|lo|>": 50336,
71
+ "<|lt|>": 50293,
72
+ "<|lv|>": 50301,
73
+ "<|mg|>": 50349,
74
+ "<|mi|>": 50295,
75
+ "<|mk|>": 50308,
76
+ "<|ml|>": 50296,
77
+ "<|mn|>": 50314,
78
+ "<|mr|>": 50320,
79
+ "<|ms|>": 50282,
80
+ "<|mt|>": 50343,
81
+ "<|my|>": 50346,
82
+ "<|ne|>": 50313,
83
+ "<|nl|>": 50271,
84
+ "<|nn|>": 50342,
85
+ "<|no|>": 50288,
86
+ "<|oc|>": 50328,
87
+ "<|pa|>": 50321,
88
+ "<|pl|>": 50269,
89
+ "<|ps|>": 50340,
90
+ "<|pt|>": 50267,
91
+ "<|ro|>": 50284,
92
+ "<|ru|>": 50263,
93
+ "<|sa|>": 50344,
94
+ "<|sd|>": 50332,
95
+ "<|si|>": 50322,
96
+ "<|sk|>": 50298,
97
+ "<|sl|>": 50305,
98
+ "<|sn|>": 50324,
99
+ "<|so|>": 50326,
100
+ "<|sq|>": 50317,
101
+ "<|sr|>": 50303,
102
+ "<|su|>": 50357,
103
+ "<|sv|>": 50273,
104
+ "<|sw|>": 50318,
105
+ "<|ta|>": 50287,
106
+ "<|te|>": 50299,
107
+ "<|tg|>": 50331,
108
+ "<|th|>": 50289,
109
+ "<|tk|>": 50341,
110
+ "<|tl|>": 50348,
111
+ "<|tr|>": 50268,
112
+ "<|tt|>": 50351,
113
+ "<|uk|>": 50280,
114
+ "<|ur|>": 50290,
115
+ "<|uz|>": 50337,
116
+ "<|vi|>": 50278,
117
+ "<|yi|>": 50335,
118
+ "<|yo|>": 50325,
119
+ "<|zh|>": 50260
120
+ },
121
+ "max_initial_timestamp_index": 1,
122
+ "max_length": 448,
123
+ "no_timestamps_token_id": 50363,
124
+ "pad_token_id": 50257,
125
+ "return_timestamps": false,
126
+ "suppress_tokens": [
127
+ 1,
128
+ 2,
129
+ 7,
130
+ 8,
131
+ 9,
132
+ 10,
133
+ 14,
134
+ 25,
135
+ 26,
136
+ 27,
137
+ 28,
138
+ 29,
139
+ 31,
140
+ 58,
141
+ 59,
142
+ 60,
143
+ 61,
144
+ 62,
145
+ 63,
146
+ 90,
147
+ 91,
148
+ 92,
149
+ 93,
150
+ 359,
151
+ 503,
152
+ 522,
153
+ 542,
154
+ 873,
155
+ 893,
156
+ 902,
157
+ 918,
158
+ 922,
159
+ 931,
160
+ 1350,
161
+ 1853,
162
+ 1982,
163
+ 2460,
164
+ 2627,
165
+ 3246,
166
+ 3253,
167
+ 3268,
168
+ 3536,
169
+ 3846,
170
+ 3961,
171
+ 4183,
172
+ 4667,
173
+ 6585,
174
+ 6647,
175
+ 7273,
176
+ 9061,
177
+ 9383,
178
+ 10428,
179
+ 10929,
180
+ 11938,
181
+ 12033,
182
+ 12331,
183
+ 12562,
184
+ 13793,
185
+ 14157,
186
+ 14635,
187
+ 15265,
188
+ 15618,
189
+ 16553,
190
+ 16604,
191
+ 18362,
192
+ 18956,
193
+ 20075,
194
+ 21675,
195
+ 22520,
196
+ 26130,
197
+ 26161,
198
+ 26435,
199
+ 28279,
200
+ 29464,
201
+ 31650,
202
+ 32302,
203
+ 32470,
204
+ 36865,
205
+ 42863,
206
+ 47425,
207
+ 49870,
208
+ 50254,
209
+ 50258,
210
+ 50358,
211
+ 50359,
212
+ 50360,
213
+ 50361,
214
+ 50362
215
+ ],
216
+ "task_to_id": {
217
+ "transcribe": 50359,
218
+ "translate": 50358
219
+ },
220
+ "transformers_version": "4.28.1"
221
+ }
checkpoint-3000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fb9bee0814e71240db234cdad4f2b2e1ef4d15e5d5ab694db8e5519904d02f95
3
+ size 1934161093
checkpoint-3000/preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 80,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
checkpoint-3000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:525d4825c016f7499f22ef7e4aa8c625a58aefcba42f778c7f3889a2697ccedd
3
+ size 967102601
checkpoint-3000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4b25041960debb0a29d2ecc74d56cc0defdc112bd8474a9897fa7523f532be2
3
+ size 14575
checkpoint-3000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5c150d8ede7f2d7791b32993758b1ba894deff3f64a7b005cb0a978a15226ff
3
+ size 627
checkpoint-3000/trainer_state.json ADDED
@@ -0,0 +1,763 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 33.16084938138103,
3
+ "best_model_checkpoint": "./whisper-small-ko-custom-1000h\\checkpoint-1000",
4
+ "epoch": 0.02007870853746687,
5
+ "global_step": 3000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 2.5e-06,
13
+ "loss": 3.5773,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 5e-06,
19
+ "loss": 1.4784,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.0,
24
+ "learning_rate": 7.500000000000001e-06,
25
+ "loss": 0.753,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.0,
30
+ "learning_rate": 1e-05,
31
+ "loss": 0.5502,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 0.0,
36
+ "learning_rate": 9.948979591836737e-06,
37
+ "loss": 0.3562,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 0.0,
42
+ "learning_rate": 9.89795918367347e-06,
43
+ "loss": 0.2842,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 0.0,
48
+ "learning_rate": 9.846938775510205e-06,
49
+ "loss": 0.2484,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 0.0,
54
+ "learning_rate": 9.795918367346939e-06,
55
+ "loss": 0.2421,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 0.0,
60
+ "learning_rate": 9.744897959183674e-06,
61
+ "loss": 0.2403,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 0.0,
66
+ "learning_rate": 9.693877551020408e-06,
67
+ "loss": 0.2222,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 0.0,
72
+ "learning_rate": 9.642857142857144e-06,
73
+ "loss": 0.202,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 0.0,
78
+ "learning_rate": 9.591836734693878e-06,
79
+ "loss": 0.2197,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 0.0,
84
+ "learning_rate": 9.540816326530612e-06,
85
+ "loss": 0.2045,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 0.0,
90
+ "learning_rate": 9.489795918367348e-06,
91
+ "loss": 0.2275,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 0.0,
96
+ "learning_rate": 9.438775510204082e-06,
97
+ "loss": 0.2023,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 0.0,
102
+ "learning_rate": 9.387755102040818e-06,
103
+ "loss": 0.2212,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 0.0,
108
+ "learning_rate": 9.336734693877552e-06,
109
+ "loss": 0.211,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 0.0,
114
+ "learning_rate": 9.285714285714288e-06,
115
+ "loss": 0.2041,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 0.0,
120
+ "learning_rate": 9.234693877551022e-06,
121
+ "loss": 0.2215,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 0.0,
126
+ "learning_rate": 9.183673469387756e-06,
127
+ "loss": 0.2012,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 0.0,
132
+ "learning_rate": 9.13265306122449e-06,
133
+ "loss": 0.1932,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 0.0,
138
+ "learning_rate": 9.081632653061225e-06,
139
+ "loss": 0.2087,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 0.0,
144
+ "learning_rate": 9.03061224489796e-06,
145
+ "loss": 0.1908,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 0.0,
150
+ "learning_rate": 8.979591836734695e-06,
151
+ "loss": 0.2019,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 0.0,
156
+ "learning_rate": 8.92857142857143e-06,
157
+ "loss": 0.2097,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 0.0,
162
+ "learning_rate": 8.877551020408163e-06,
163
+ "loss": 0.2105,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 0.0,
168
+ "learning_rate": 8.826530612244899e-06,
169
+ "loss": 0.1846,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 0.0,
174
+ "learning_rate": 8.775510204081633e-06,
175
+ "loss": 0.1748,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 0.0,
180
+ "learning_rate": 8.724489795918369e-06,
181
+ "loss": 0.1761,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 0.01,
186
+ "learning_rate": 8.673469387755103e-06,
187
+ "loss": 0.1982,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 0.01,
192
+ "learning_rate": 8.622448979591837e-06,
193
+ "loss": 0.1926,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 0.01,
198
+ "learning_rate": 8.571428571428571e-06,
199
+ "loss": 0.1732,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 0.01,
204
+ "learning_rate": 8.520408163265307e-06,
205
+ "loss": 0.1841,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 0.01,
210
+ "learning_rate": 8.469387755102042e-06,
211
+ "loss": 0.1699,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 0.01,
216
+ "learning_rate": 8.418367346938776e-06,
217
+ "loss": 0.2131,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 0.01,
222
+ "learning_rate": 8.36734693877551e-06,
223
+ "loss": 0.1828,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 0.01,
228
+ "learning_rate": 8.316326530612246e-06,
229
+ "loss": 0.1796,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 0.01,
234
+ "learning_rate": 8.26530612244898e-06,
235
+ "loss": 0.1909,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 0.01,
240
+ "learning_rate": 8.214285714285714e-06,
241
+ "loss": 0.2066,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 0.01,
246
+ "learning_rate": 8.16326530612245e-06,
247
+ "loss": 0.1913,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 0.01,
252
+ "eval_loss": 0.17891158163547516,
253
+ "eval_runtime": 14046.7097,
254
+ "eval_samples_per_second": 3.473,
255
+ "eval_steps_per_second": 0.434,
256
+ "eval_wer": 33.16084938138103,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 0.01,
261
+ "learning_rate": 8.112244897959184e-06,
262
+ "loss": 0.1426,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 0.01,
267
+ "learning_rate": 8.06122448979592e-06,
268
+ "loss": 0.1976,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 0.01,
273
+ "learning_rate": 8.010204081632654e-06,
274
+ "loss": 0.1677,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 0.01,
279
+ "learning_rate": 7.959183673469388e-06,
280
+ "loss": 0.1633,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 0.01,
285
+ "learning_rate": 7.908163265306124e-06,
286
+ "loss": 0.1994,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 0.01,
291
+ "learning_rate": 7.857142857142858e-06,
292
+ "loss": 0.162,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 0.01,
297
+ "learning_rate": 7.806122448979593e-06,
298
+ "loss": 0.1779,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 0.01,
303
+ "learning_rate": 7.755102040816327e-06,
304
+ "loss": 0.1829,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 0.01,
309
+ "learning_rate": 7.704081632653061e-06,
310
+ "loss": 0.1933,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 0.01,
315
+ "learning_rate": 7.653061224489796e-06,
316
+ "loss": 0.1606,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 0.01,
321
+ "learning_rate": 7.602040816326531e-06,
322
+ "loss": 0.1785,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 0.01,
327
+ "learning_rate": 7.551020408163265e-06,
328
+ "loss": 0.1802,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 0.01,
333
+ "learning_rate": 7.500000000000001e-06,
334
+ "loss": 0.17,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 0.01,
339
+ "learning_rate": 7.448979591836736e-06,
340
+ "loss": 0.1679,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 0.01,
345
+ "learning_rate": 7.39795918367347e-06,
346
+ "loss": 0.157,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 0.01,
351
+ "learning_rate": 7.346938775510205e-06,
352
+ "loss": 0.161,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 0.01,
357
+ "learning_rate": 7.295918367346939e-06,
358
+ "loss": 0.1494,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 0.01,
363
+ "learning_rate": 7.244897959183675e-06,
364
+ "loss": 0.1799,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 0.01,
369
+ "learning_rate": 7.193877551020409e-06,
370
+ "loss": 0.1668,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 0.01,
375
+ "learning_rate": 7.1428571428571436e-06,
376
+ "loss": 0.1798,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 0.01,
381
+ "learning_rate": 7.091836734693878e-06,
382
+ "loss": 0.1476,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 0.01,
387
+ "learning_rate": 7.0408163265306125e-06,
388
+ "loss": 0.1779,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 0.01,
393
+ "learning_rate": 6.989795918367348e-06,
394
+ "loss": 0.1795,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 0.01,
399
+ "learning_rate": 6.938775510204082e-06,
400
+ "loss": 0.1642,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 0.01,
405
+ "learning_rate": 6.887755102040817e-06,
406
+ "loss": 0.1239,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 0.01,
411
+ "learning_rate": 6.836734693877551e-06,
412
+ "loss": 0.2013,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 0.01,
417
+ "learning_rate": 6.785714285714287e-06,
418
+ "loss": 0.1646,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 0.01,
423
+ "learning_rate": 6.734693877551021e-06,
424
+ "loss": 0.1339,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 0.01,
429
+ "learning_rate": 6.683673469387756e-06,
430
+ "loss": 0.1814,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 0.01,
435
+ "learning_rate": 6.63265306122449e-06,
436
+ "loss": 0.1302,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 0.01,
441
+ "learning_rate": 6.581632653061225e-06,
442
+ "loss": 0.1324,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 0.01,
447
+ "learning_rate": 6.530612244897959e-06,
448
+ "loss": 0.1384,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 0.01,
453
+ "learning_rate": 6.4795918367346946e-06,
454
+ "loss": 0.1628,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 0.01,
459
+ "learning_rate": 6.4285714285714295e-06,
460
+ "loss": 0.1712,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 0.01,
465
+ "learning_rate": 6.3775510204081635e-06,
466
+ "loss": 0.153,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 0.01,
471
+ "learning_rate": 6.326530612244899e-06,
472
+ "loss": 0.1588,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 0.01,
477
+ "learning_rate": 6.275510204081633e-06,
478
+ "loss": 0.1579,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 0.01,
483
+ "learning_rate": 6.224489795918368e-06,
484
+ "loss": 0.2014,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 0.01,
489
+ "learning_rate": 6.173469387755102e-06,
490
+ "loss": 0.153,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 0.01,
495
+ "learning_rate": 6.122448979591837e-06,
496
+ "loss": 0.141,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 0.01,
501
+ "eval_loss": 0.153132826089859,
502
+ "eval_runtime": 14050.6821,
503
+ "eval_samples_per_second": 3.472,
504
+ "eval_steps_per_second": 0.434,
505
+ "eval_wer": 33.639439546838894,
506
+ "step": 2000
507
+ },
508
+ {
509
+ "epoch": 0.01,
510
+ "learning_rate": 6.071428571428571e-06,
511
+ "loss": 0.1496,
512
+ "step": 2025
513
+ },
514
+ {
515
+ "epoch": 0.01,
516
+ "learning_rate": 6.020408163265307e-06,
517
+ "loss": 0.1456,
518
+ "step": 2050
519
+ },
520
+ {
521
+ "epoch": 0.01,
522
+ "learning_rate": 5.969387755102042e-06,
523
+ "loss": 0.1668,
524
+ "step": 2075
525
+ },
526
+ {
527
+ "epoch": 0.01,
528
+ "learning_rate": 5.918367346938776e-06,
529
+ "loss": 0.1737,
530
+ "step": 2100
531
+ },
532
+ {
533
+ "epoch": 0.01,
534
+ "learning_rate": 5.867346938775511e-06,
535
+ "loss": 0.1378,
536
+ "step": 2125
537
+ },
538
+ {
539
+ "epoch": 0.01,
540
+ "learning_rate": 5.816326530612246e-06,
541
+ "loss": 0.159,
542
+ "step": 2150
543
+ },
544
+ {
545
+ "epoch": 0.01,
546
+ "learning_rate": 5.7653061224489805e-06,
547
+ "loss": 0.1262,
548
+ "step": 2175
549
+ },
550
+ {
551
+ "epoch": 0.01,
552
+ "learning_rate": 5.7142857142857145e-06,
553
+ "loss": 0.1801,
554
+ "step": 2200
555
+ },
556
+ {
557
+ "epoch": 0.01,
558
+ "learning_rate": 5.663265306122449e-06,
559
+ "loss": 0.1221,
560
+ "step": 2225
561
+ },
562
+ {
563
+ "epoch": 0.02,
564
+ "learning_rate": 5.6122448979591834e-06,
565
+ "loss": 0.1261,
566
+ "step": 2250
567
+ },
568
+ {
569
+ "epoch": 0.02,
570
+ "learning_rate": 5.561224489795919e-06,
571
+ "loss": 0.1584,
572
+ "step": 2275
573
+ },
574
+ {
575
+ "epoch": 0.02,
576
+ "learning_rate": 5.510204081632653e-06,
577
+ "loss": 0.1534,
578
+ "step": 2300
579
+ },
580
+ {
581
+ "epoch": 0.02,
582
+ "learning_rate": 5.459183673469388e-06,
583
+ "loss": 0.1572,
584
+ "step": 2325
585
+ },
586
+ {
587
+ "epoch": 0.02,
588
+ "learning_rate": 5.408163265306123e-06,
589
+ "loss": 0.1091,
590
+ "step": 2350
591
+ },
592
+ {
593
+ "epoch": 0.02,
594
+ "learning_rate": 5.357142857142857e-06,
595
+ "loss": 0.1719,
596
+ "step": 2375
597
+ },
598
+ {
599
+ "epoch": 0.02,
600
+ "learning_rate": 5.306122448979593e-06,
601
+ "loss": 0.1817,
602
+ "step": 2400
603
+ },
604
+ {
605
+ "epoch": 0.02,
606
+ "learning_rate": 5.255102040816327e-06,
607
+ "loss": 0.1407,
608
+ "step": 2425
609
+ },
610
+ {
611
+ "epoch": 0.02,
612
+ "learning_rate": 5.204081632653062e-06,
613
+ "loss": 0.1418,
614
+ "step": 2450
615
+ },
616
+ {
617
+ "epoch": 0.02,
618
+ "learning_rate": 5.153061224489796e-06,
619
+ "loss": 0.1489,
620
+ "step": 2475
621
+ },
622
+ {
623
+ "epoch": 0.02,
624
+ "learning_rate": 5.1020408163265315e-06,
625
+ "loss": 0.1542,
626
+ "step": 2500
627
+ },
628
+ {
629
+ "epoch": 0.02,
630
+ "learning_rate": 5.0510204081632655e-06,
631
+ "loss": 0.1643,
632
+ "step": 2525
633
+ },
634
+ {
635
+ "epoch": 0.02,
636
+ "learning_rate": 5e-06,
637
+ "loss": 0.1275,
638
+ "step": 2550
639
+ },
640
+ {
641
+ "epoch": 0.02,
642
+ "learning_rate": 4.948979591836735e-06,
643
+ "loss": 0.1409,
644
+ "step": 2575
645
+ },
646
+ {
647
+ "epoch": 0.02,
648
+ "learning_rate": 4.897959183673469e-06,
649
+ "loss": 0.1463,
650
+ "step": 2600
651
+ },
652
+ {
653
+ "epoch": 0.02,
654
+ "learning_rate": 4.846938775510204e-06,
655
+ "loss": 0.1329,
656
+ "step": 2625
657
+ },
658
+ {
659
+ "epoch": 0.02,
660
+ "learning_rate": 4.795918367346939e-06,
661
+ "loss": 0.1499,
662
+ "step": 2650
663
+ },
664
+ {
665
+ "epoch": 0.02,
666
+ "learning_rate": 4.744897959183674e-06,
667
+ "loss": 0.121,
668
+ "step": 2675
669
+ },
670
+ {
671
+ "epoch": 0.02,
672
+ "learning_rate": 4.693877551020409e-06,
673
+ "loss": 0.1134,
674
+ "step": 2700
675
+ },
676
+ {
677
+ "epoch": 0.02,
678
+ "learning_rate": 4.642857142857144e-06,
679
+ "loss": 0.1321,
680
+ "step": 2725
681
+ },
682
+ {
683
+ "epoch": 0.02,
684
+ "learning_rate": 4.591836734693878e-06,
685
+ "loss": 0.1424,
686
+ "step": 2750
687
+ },
688
+ {
689
+ "epoch": 0.02,
690
+ "learning_rate": 4.540816326530613e-06,
691
+ "loss": 0.1478,
692
+ "step": 2775
693
+ },
694
+ {
695
+ "epoch": 0.02,
696
+ "learning_rate": 4.489795918367348e-06,
697
+ "loss": 0.1289,
698
+ "step": 2800
699
+ },
700
+ {
701
+ "epoch": 0.02,
702
+ "learning_rate": 4.438775510204082e-06,
703
+ "loss": 0.1426,
704
+ "step": 2825
705
+ },
706
+ {
707
+ "epoch": 0.02,
708
+ "learning_rate": 4.3877551020408165e-06,
709
+ "loss": 0.1541,
710
+ "step": 2850
711
+ },
712
+ {
713
+ "epoch": 0.02,
714
+ "learning_rate": 4.336734693877551e-06,
715
+ "loss": 0.1242,
716
+ "step": 2875
717
+ },
718
+ {
719
+ "epoch": 0.02,
720
+ "learning_rate": 4.2857142857142855e-06,
721
+ "loss": 0.1589,
722
+ "step": 2900
723
+ },
724
+ {
725
+ "epoch": 0.02,
726
+ "learning_rate": 4.234693877551021e-06,
727
+ "loss": 0.1425,
728
+ "step": 2925
729
+ },
730
+ {
731
+ "epoch": 0.02,
732
+ "learning_rate": 4.183673469387755e-06,
733
+ "loss": 0.1521,
734
+ "step": 2950
735
+ },
736
+ {
737
+ "epoch": 0.02,
738
+ "learning_rate": 4.13265306122449e-06,
739
+ "loss": 0.1434,
740
+ "step": 2975
741
+ },
742
+ {
743
+ "epoch": 0.02,
744
+ "learning_rate": 4.081632653061225e-06,
745
+ "loss": 0.1477,
746
+ "step": 3000
747
+ },
748
+ {
749
+ "epoch": 0.02,
750
+ "eval_loss": 0.13808605074882507,
751
+ "eval_runtime": 13604.3241,
752
+ "eval_samples_per_second": 3.586,
753
+ "eval_steps_per_second": 0.448,
754
+ "eval_wer": 37.97294043601127,
755
+ "step": 3000
756
+ }
757
+ ],
758
+ "max_steps": 5000,
759
+ "num_train_epochs": 1,
760
+ "total_flos": 1.385209921536e+19,
761
+ "trial_name": null,
762
+ "trial_params": null
763
+ }
checkpoint-3000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe99781f928141fde56544c833b4f8a8a1100a885088c1e45942567c2ed1ebfc
3
+ size 3771
checkpoint-4000/config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "model_type": "whisper",
41
+ "num_hidden_layers": 12,
42
+ "num_mel_bins": 80,
43
+ "pad_token_id": 50257,
44
+ "scale_embedding": false,
45
+ "suppress_tokens": [],
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.28.1",
48
+ "use_cache": true,
49
+ "use_weighted_layer_sum": false,
50
+ "vocab_size": 51865
51
+ }
checkpoint-4000/generation_config.json ADDED
@@ -0,0 +1,221 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "begin_suppress_tokens": [
3
+ 220,
4
+ 50257
5
+ ],
6
+ "bos_token_id": 50257,
7
+ "decoder_start_token_id": 50258,
8
+ "eos_token_id": 50257,
9
+ "forced_decoder_ids": [
10
+ [
11
+ 1,
12
+ null
13
+ ],
14
+ [
15
+ 2,
16
+ 50359
17
+ ]
18
+ ],
19
+ "is_multilingual": true,
20
+ "lang_to_id": {
21
+ "<|af|>": 50327,
22
+ "<|am|>": 50334,
23
+ "<|ar|>": 50272,
24
+ "<|as|>": 50350,
25
+ "<|az|>": 50304,
26
+ "<|ba|>": 50355,
27
+ "<|be|>": 50330,
28
+ "<|bg|>": 50292,
29
+ "<|bn|>": 50302,
30
+ "<|bo|>": 50347,
31
+ "<|br|>": 50309,
32
+ "<|bs|>": 50315,
33
+ "<|ca|>": 50270,
34
+ "<|cs|>": 50283,
35
+ "<|cy|>": 50297,
36
+ "<|da|>": 50285,
37
+ "<|de|>": 50261,
38
+ "<|el|>": 50281,
39
+ "<|en|>": 50259,
40
+ "<|es|>": 50262,
41
+ "<|et|>": 50307,
42
+ "<|eu|>": 50310,
43
+ "<|fa|>": 50300,
44
+ "<|fi|>": 50277,
45
+ "<|fo|>": 50338,
46
+ "<|fr|>": 50265,
47
+ "<|gl|>": 50319,
48
+ "<|gu|>": 50333,
49
+ "<|haw|>": 50352,
50
+ "<|ha|>": 50354,
51
+ "<|he|>": 50279,
52
+ "<|hi|>": 50276,
53
+ "<|hr|>": 50291,
54
+ "<|ht|>": 50339,
55
+ "<|hu|>": 50286,
56
+ "<|hy|>": 50312,
57
+ "<|id|>": 50275,
58
+ "<|is|>": 50311,
59
+ "<|it|>": 50274,
60
+ "<|ja|>": 50266,
61
+ "<|jw|>": 50356,
62
+ "<|ka|>": 50329,
63
+ "<|kk|>": 50316,
64
+ "<|km|>": 50323,
65
+ "<|kn|>": 50306,
66
+ "<|ko|>": 50264,
67
+ "<|la|>": 50294,
68
+ "<|lb|>": 50345,
69
+ "<|ln|>": 50353,
70
+ "<|lo|>": 50336,
71
+ "<|lt|>": 50293,
72
+ "<|lv|>": 50301,
73
+ "<|mg|>": 50349,
74
+ "<|mi|>": 50295,
75
+ "<|mk|>": 50308,
76
+ "<|ml|>": 50296,
77
+ "<|mn|>": 50314,
78
+ "<|mr|>": 50320,
79
+ "<|ms|>": 50282,
80
+ "<|mt|>": 50343,
81
+ "<|my|>": 50346,
82
+ "<|ne|>": 50313,
83
+ "<|nl|>": 50271,
84
+ "<|nn|>": 50342,
85
+ "<|no|>": 50288,
86
+ "<|oc|>": 50328,
87
+ "<|pa|>": 50321,
88
+ "<|pl|>": 50269,
89
+ "<|ps|>": 50340,
90
+ "<|pt|>": 50267,
91
+ "<|ro|>": 50284,
92
+ "<|ru|>": 50263,
93
+ "<|sa|>": 50344,
94
+ "<|sd|>": 50332,
95
+ "<|si|>": 50322,
96
+ "<|sk|>": 50298,
97
+ "<|sl|>": 50305,
98
+ "<|sn|>": 50324,
99
+ "<|so|>": 50326,
100
+ "<|sq|>": 50317,
101
+ "<|sr|>": 50303,
102
+ "<|su|>": 50357,
103
+ "<|sv|>": 50273,
104
+ "<|sw|>": 50318,
105
+ "<|ta|>": 50287,
106
+ "<|te|>": 50299,
107
+ "<|tg|>": 50331,
108
+ "<|th|>": 50289,
109
+ "<|tk|>": 50341,
110
+ "<|tl|>": 50348,
111
+ "<|tr|>": 50268,
112
+ "<|tt|>": 50351,
113
+ "<|uk|>": 50280,
114
+ "<|ur|>": 50290,
115
+ "<|uz|>": 50337,
116
+ "<|vi|>": 50278,
117
+ "<|yi|>": 50335,
118
+ "<|yo|>": 50325,
119
+ "<|zh|>": 50260
120
+ },
121
+ "max_initial_timestamp_index": 1,
122
+ "max_length": 448,
123
+ "no_timestamps_token_id": 50363,
124
+ "pad_token_id": 50257,
125
+ "return_timestamps": false,
126
+ "suppress_tokens": [
127
+ 1,
128
+ 2,
129
+ 7,
130
+ 8,
131
+ 9,
132
+ 10,
133
+ 14,
134
+ 25,
135
+ 26,
136
+ 27,
137
+ 28,
138
+ 29,
139
+ 31,
140
+ 58,
141
+ 59,
142
+ 60,
143
+ 61,
144
+ 62,
145
+ 63,
146
+ 90,
147
+ 91,
148
+ 92,
149
+ 93,
150
+ 359,
151
+ 503,
152
+ 522,
153
+ 542,
154
+ 873,
155
+ 893,
156
+ 902,
157
+ 918,
158
+ 922,
159
+ 931,
160
+ 1350,
161
+ 1853,
162
+ 1982,
163
+ 2460,
164
+ 2627,
165
+ 3246,
166
+ 3253,
167
+ 3268,
168
+ 3536,
169
+ 3846,
170
+ 3961,
171
+ 4183,
172
+ 4667,
173
+ 6585,
174
+ 6647,
175
+ 7273,
176
+ 9061,
177
+ 9383,
178
+ 10428,
179
+ 10929,
180
+ 11938,
181
+ 12033,
182
+ 12331,
183
+ 12562,
184
+ 13793,
185
+ 14157,
186
+ 14635,
187
+ 15265,
188
+ 15618,
189
+ 16553,
190
+ 16604,
191
+ 18362,
192
+ 18956,
193
+ 20075,
194
+ 21675,
195
+ 22520,
196
+ 26130,
197
+ 26161,
198
+ 26435,
199
+ 28279,
200
+ 29464,
201
+ 31650,
202
+ 32302,
203
+ 32470,
204
+ 36865,
205
+ 42863,
206
+ 47425,
207
+ 49870,
208
+ 50254,
209
+ 50258,
210
+ 50358,
211
+ 50359,
212
+ 50360,
213
+ 50361,
214
+ 50362
215
+ ],
216
+ "task_to_id": {
217
+ "transcribe": 50359,
218
+ "translate": 50358
219
+ },
220
+ "transformers_version": "4.28.1"
221
+ }
checkpoint-4000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a26cab3df84538f714ad70afd6bc3a1af5d994d93b937cf75579156cacc4a38f
3
+ size 1934161093
checkpoint-4000/preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 80,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
checkpoint-4000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1c81a3245a401b82e3154f253669ae1303cc2c3974d000ef9480962976091773
3
+ size 967102601
checkpoint-4000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:daf575b93c611c8c17f87b755d6cf4c2331aaf24fb9a1af4cde6967f79fbfc70
3
+ size 14639
checkpoint-4000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5a3a8b8efde7a7242400115171269cf0af16b91c7bb89d700ea07585d51b5364
3
+ size 627
checkpoint-4000/trainer_state.json ADDED
@@ -0,0 +1,1012 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 33.16084938138103,
3
+ "best_model_checkpoint": "./whisper-small-ko-custom-1000h\\checkpoint-1000",
4
+ "epoch": 0.02677161138328916,
5
+ "global_step": 4000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 2.5e-06,
13
+ "loss": 3.5773,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 5e-06,
19
+ "loss": 1.4784,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.0,
24
+ "learning_rate": 7.500000000000001e-06,
25
+ "loss": 0.753,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.0,
30
+ "learning_rate": 1e-05,
31
+ "loss": 0.5502,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 0.0,
36
+ "learning_rate": 9.948979591836737e-06,
37
+ "loss": 0.3562,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 0.0,
42
+ "learning_rate": 9.89795918367347e-06,
43
+ "loss": 0.2842,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 0.0,
48
+ "learning_rate": 9.846938775510205e-06,
49
+ "loss": 0.2484,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 0.0,
54
+ "learning_rate": 9.795918367346939e-06,
55
+ "loss": 0.2421,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 0.0,
60
+ "learning_rate": 9.744897959183674e-06,
61
+ "loss": 0.2403,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 0.0,
66
+ "learning_rate": 9.693877551020408e-06,
67
+ "loss": 0.2222,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 0.0,
72
+ "learning_rate": 9.642857142857144e-06,
73
+ "loss": 0.202,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 0.0,
78
+ "learning_rate": 9.591836734693878e-06,
79
+ "loss": 0.2197,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 0.0,
84
+ "learning_rate": 9.540816326530612e-06,
85
+ "loss": 0.2045,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 0.0,
90
+ "learning_rate": 9.489795918367348e-06,
91
+ "loss": 0.2275,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 0.0,
96
+ "learning_rate": 9.438775510204082e-06,
97
+ "loss": 0.2023,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 0.0,
102
+ "learning_rate": 9.387755102040818e-06,
103
+ "loss": 0.2212,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 0.0,
108
+ "learning_rate": 9.336734693877552e-06,
109
+ "loss": 0.211,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 0.0,
114
+ "learning_rate": 9.285714285714288e-06,
115
+ "loss": 0.2041,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 0.0,
120
+ "learning_rate": 9.234693877551022e-06,
121
+ "loss": 0.2215,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 0.0,
126
+ "learning_rate": 9.183673469387756e-06,
127
+ "loss": 0.2012,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 0.0,
132
+ "learning_rate": 9.13265306122449e-06,
133
+ "loss": 0.1932,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 0.0,
138
+ "learning_rate": 9.081632653061225e-06,
139
+ "loss": 0.2087,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 0.0,
144
+ "learning_rate": 9.03061224489796e-06,
145
+ "loss": 0.1908,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 0.0,
150
+ "learning_rate": 8.979591836734695e-06,
151
+ "loss": 0.2019,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 0.0,
156
+ "learning_rate": 8.92857142857143e-06,
157
+ "loss": 0.2097,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 0.0,
162
+ "learning_rate": 8.877551020408163e-06,
163
+ "loss": 0.2105,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 0.0,
168
+ "learning_rate": 8.826530612244899e-06,
169
+ "loss": 0.1846,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 0.0,
174
+ "learning_rate": 8.775510204081633e-06,
175
+ "loss": 0.1748,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 0.0,
180
+ "learning_rate": 8.724489795918369e-06,
181
+ "loss": 0.1761,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 0.01,
186
+ "learning_rate": 8.673469387755103e-06,
187
+ "loss": 0.1982,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 0.01,
192
+ "learning_rate": 8.622448979591837e-06,
193
+ "loss": 0.1926,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 0.01,
198
+ "learning_rate": 8.571428571428571e-06,
199
+ "loss": 0.1732,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 0.01,
204
+ "learning_rate": 8.520408163265307e-06,
205
+ "loss": 0.1841,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 0.01,
210
+ "learning_rate": 8.469387755102042e-06,
211
+ "loss": 0.1699,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 0.01,
216
+ "learning_rate": 8.418367346938776e-06,
217
+ "loss": 0.2131,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 0.01,
222
+ "learning_rate": 8.36734693877551e-06,
223
+ "loss": 0.1828,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 0.01,
228
+ "learning_rate": 8.316326530612246e-06,
229
+ "loss": 0.1796,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 0.01,
234
+ "learning_rate": 8.26530612244898e-06,
235
+ "loss": 0.1909,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 0.01,
240
+ "learning_rate": 8.214285714285714e-06,
241
+ "loss": 0.2066,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 0.01,
246
+ "learning_rate": 8.16326530612245e-06,
247
+ "loss": 0.1913,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 0.01,
252
+ "eval_loss": 0.17891158163547516,
253
+ "eval_runtime": 14046.7097,
254
+ "eval_samples_per_second": 3.473,
255
+ "eval_steps_per_second": 0.434,
256
+ "eval_wer": 33.16084938138103,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 0.01,
261
+ "learning_rate": 8.112244897959184e-06,
262
+ "loss": 0.1426,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 0.01,
267
+ "learning_rate": 8.06122448979592e-06,
268
+ "loss": 0.1976,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 0.01,
273
+ "learning_rate": 8.010204081632654e-06,
274
+ "loss": 0.1677,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 0.01,
279
+ "learning_rate": 7.959183673469388e-06,
280
+ "loss": 0.1633,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 0.01,
285
+ "learning_rate": 7.908163265306124e-06,
286
+ "loss": 0.1994,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 0.01,
291
+ "learning_rate": 7.857142857142858e-06,
292
+ "loss": 0.162,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 0.01,
297
+ "learning_rate": 7.806122448979593e-06,
298
+ "loss": 0.1779,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 0.01,
303
+ "learning_rate": 7.755102040816327e-06,
304
+ "loss": 0.1829,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 0.01,
309
+ "learning_rate": 7.704081632653061e-06,
310
+ "loss": 0.1933,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 0.01,
315
+ "learning_rate": 7.653061224489796e-06,
316
+ "loss": 0.1606,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 0.01,
321
+ "learning_rate": 7.602040816326531e-06,
322
+ "loss": 0.1785,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 0.01,
327
+ "learning_rate": 7.551020408163265e-06,
328
+ "loss": 0.1802,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 0.01,
333
+ "learning_rate": 7.500000000000001e-06,
334
+ "loss": 0.17,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 0.01,
339
+ "learning_rate": 7.448979591836736e-06,
340
+ "loss": 0.1679,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 0.01,
345
+ "learning_rate": 7.39795918367347e-06,
346
+ "loss": 0.157,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 0.01,
351
+ "learning_rate": 7.346938775510205e-06,
352
+ "loss": 0.161,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 0.01,
357
+ "learning_rate": 7.295918367346939e-06,
358
+ "loss": 0.1494,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 0.01,
363
+ "learning_rate": 7.244897959183675e-06,
364
+ "loss": 0.1799,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 0.01,
369
+ "learning_rate": 7.193877551020409e-06,
370
+ "loss": 0.1668,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 0.01,
375
+ "learning_rate": 7.1428571428571436e-06,
376
+ "loss": 0.1798,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 0.01,
381
+ "learning_rate": 7.091836734693878e-06,
382
+ "loss": 0.1476,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 0.01,
387
+ "learning_rate": 7.0408163265306125e-06,
388
+ "loss": 0.1779,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 0.01,
393
+ "learning_rate": 6.989795918367348e-06,
394
+ "loss": 0.1795,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 0.01,
399
+ "learning_rate": 6.938775510204082e-06,
400
+ "loss": 0.1642,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 0.01,
405
+ "learning_rate": 6.887755102040817e-06,
406
+ "loss": 0.1239,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 0.01,
411
+ "learning_rate": 6.836734693877551e-06,
412
+ "loss": 0.2013,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 0.01,
417
+ "learning_rate": 6.785714285714287e-06,
418
+ "loss": 0.1646,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 0.01,
423
+ "learning_rate": 6.734693877551021e-06,
424
+ "loss": 0.1339,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 0.01,
429
+ "learning_rate": 6.683673469387756e-06,
430
+ "loss": 0.1814,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 0.01,
435
+ "learning_rate": 6.63265306122449e-06,
436
+ "loss": 0.1302,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 0.01,
441
+ "learning_rate": 6.581632653061225e-06,
442
+ "loss": 0.1324,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 0.01,
447
+ "learning_rate": 6.530612244897959e-06,
448
+ "loss": 0.1384,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 0.01,
453
+ "learning_rate": 6.4795918367346946e-06,
454
+ "loss": 0.1628,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 0.01,
459
+ "learning_rate": 6.4285714285714295e-06,
460
+ "loss": 0.1712,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 0.01,
465
+ "learning_rate": 6.3775510204081635e-06,
466
+ "loss": 0.153,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 0.01,
471
+ "learning_rate": 6.326530612244899e-06,
472
+ "loss": 0.1588,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 0.01,
477
+ "learning_rate": 6.275510204081633e-06,
478
+ "loss": 0.1579,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 0.01,
483
+ "learning_rate": 6.224489795918368e-06,
484
+ "loss": 0.2014,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 0.01,
489
+ "learning_rate": 6.173469387755102e-06,
490
+ "loss": 0.153,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 0.01,
495
+ "learning_rate": 6.122448979591837e-06,
496
+ "loss": 0.141,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 0.01,
501
+ "eval_loss": 0.153132826089859,
502
+ "eval_runtime": 14050.6821,
503
+ "eval_samples_per_second": 3.472,
504
+ "eval_steps_per_second": 0.434,
505
+ "eval_wer": 33.639439546838894,
506
+ "step": 2000
507
+ },
508
+ {
509
+ "epoch": 0.01,
510
+ "learning_rate": 6.071428571428571e-06,
511
+ "loss": 0.1496,
512
+ "step": 2025
513
+ },
514
+ {
515
+ "epoch": 0.01,
516
+ "learning_rate": 6.020408163265307e-06,
517
+ "loss": 0.1456,
518
+ "step": 2050
519
+ },
520
+ {
521
+ "epoch": 0.01,
522
+ "learning_rate": 5.969387755102042e-06,
523
+ "loss": 0.1668,
524
+ "step": 2075
525
+ },
526
+ {
527
+ "epoch": 0.01,
528
+ "learning_rate": 5.918367346938776e-06,
529
+ "loss": 0.1737,
530
+ "step": 2100
531
+ },
532
+ {
533
+ "epoch": 0.01,
534
+ "learning_rate": 5.867346938775511e-06,
535
+ "loss": 0.1378,
536
+ "step": 2125
537
+ },
538
+ {
539
+ "epoch": 0.01,
540
+ "learning_rate": 5.816326530612246e-06,
541
+ "loss": 0.159,
542
+ "step": 2150
543
+ },
544
+ {
545
+ "epoch": 0.01,
546
+ "learning_rate": 5.7653061224489805e-06,
547
+ "loss": 0.1262,
548
+ "step": 2175
549
+ },
550
+ {
551
+ "epoch": 0.01,
552
+ "learning_rate": 5.7142857142857145e-06,
553
+ "loss": 0.1801,
554
+ "step": 2200
555
+ },
556
+ {
557
+ "epoch": 0.01,
558
+ "learning_rate": 5.663265306122449e-06,
559
+ "loss": 0.1221,
560
+ "step": 2225
561
+ },
562
+ {
563
+ "epoch": 0.02,
564
+ "learning_rate": 5.6122448979591834e-06,
565
+ "loss": 0.1261,
566
+ "step": 2250
567
+ },
568
+ {
569
+ "epoch": 0.02,
570
+ "learning_rate": 5.561224489795919e-06,
571
+ "loss": 0.1584,
572
+ "step": 2275
573
+ },
574
+ {
575
+ "epoch": 0.02,
576
+ "learning_rate": 5.510204081632653e-06,
577
+ "loss": 0.1534,
578
+ "step": 2300
579
+ },
580
+ {
581
+ "epoch": 0.02,
582
+ "learning_rate": 5.459183673469388e-06,
583
+ "loss": 0.1572,
584
+ "step": 2325
585
+ },
586
+ {
587
+ "epoch": 0.02,
588
+ "learning_rate": 5.408163265306123e-06,
589
+ "loss": 0.1091,
590
+ "step": 2350
591
+ },
592
+ {
593
+ "epoch": 0.02,
594
+ "learning_rate": 5.357142857142857e-06,
595
+ "loss": 0.1719,
596
+ "step": 2375
597
+ },
598
+ {
599
+ "epoch": 0.02,
600
+ "learning_rate": 5.306122448979593e-06,
601
+ "loss": 0.1817,
602
+ "step": 2400
603
+ },
604
+ {
605
+ "epoch": 0.02,
606
+ "learning_rate": 5.255102040816327e-06,
607
+ "loss": 0.1407,
608
+ "step": 2425
609
+ },
610
+ {
611
+ "epoch": 0.02,
612
+ "learning_rate": 5.204081632653062e-06,
613
+ "loss": 0.1418,
614
+ "step": 2450
615
+ },
616
+ {
617
+ "epoch": 0.02,
618
+ "learning_rate": 5.153061224489796e-06,
619
+ "loss": 0.1489,
620
+ "step": 2475
621
+ },
622
+ {
623
+ "epoch": 0.02,
624
+ "learning_rate": 5.1020408163265315e-06,
625
+ "loss": 0.1542,
626
+ "step": 2500
627
+ },
628
+ {
629
+ "epoch": 0.02,
630
+ "learning_rate": 5.0510204081632655e-06,
631
+ "loss": 0.1643,
632
+ "step": 2525
633
+ },
634
+ {
635
+ "epoch": 0.02,
636
+ "learning_rate": 5e-06,
637
+ "loss": 0.1275,
638
+ "step": 2550
639
+ },
640
+ {
641
+ "epoch": 0.02,
642
+ "learning_rate": 4.948979591836735e-06,
643
+ "loss": 0.1409,
644
+ "step": 2575
645
+ },
646
+ {
647
+ "epoch": 0.02,
648
+ "learning_rate": 4.897959183673469e-06,
649
+ "loss": 0.1463,
650
+ "step": 2600
651
+ },
652
+ {
653
+ "epoch": 0.02,
654
+ "learning_rate": 4.846938775510204e-06,
655
+ "loss": 0.1329,
656
+ "step": 2625
657
+ },
658
+ {
659
+ "epoch": 0.02,
660
+ "learning_rate": 4.795918367346939e-06,
661
+ "loss": 0.1499,
662
+ "step": 2650
663
+ },
664
+ {
665
+ "epoch": 0.02,
666
+ "learning_rate": 4.744897959183674e-06,
667
+ "loss": 0.121,
668
+ "step": 2675
669
+ },
670
+ {
671
+ "epoch": 0.02,
672
+ "learning_rate": 4.693877551020409e-06,
673
+ "loss": 0.1134,
674
+ "step": 2700
675
+ },
676
+ {
677
+ "epoch": 0.02,
678
+ "learning_rate": 4.642857142857144e-06,
679
+ "loss": 0.1321,
680
+ "step": 2725
681
+ },
682
+ {
683
+ "epoch": 0.02,
684
+ "learning_rate": 4.591836734693878e-06,
685
+ "loss": 0.1424,
686
+ "step": 2750
687
+ },
688
+ {
689
+ "epoch": 0.02,
690
+ "learning_rate": 4.540816326530613e-06,
691
+ "loss": 0.1478,
692
+ "step": 2775
693
+ },
694
+ {
695
+ "epoch": 0.02,
696
+ "learning_rate": 4.489795918367348e-06,
697
+ "loss": 0.1289,
698
+ "step": 2800
699
+ },
700
+ {
701
+ "epoch": 0.02,
702
+ "learning_rate": 4.438775510204082e-06,
703
+ "loss": 0.1426,
704
+ "step": 2825
705
+ },
706
+ {
707
+ "epoch": 0.02,
708
+ "learning_rate": 4.3877551020408165e-06,
709
+ "loss": 0.1541,
710
+ "step": 2850
711
+ },
712
+ {
713
+ "epoch": 0.02,
714
+ "learning_rate": 4.336734693877551e-06,
715
+ "loss": 0.1242,
716
+ "step": 2875
717
+ },
718
+ {
719
+ "epoch": 0.02,
720
+ "learning_rate": 4.2857142857142855e-06,
721
+ "loss": 0.1589,
722
+ "step": 2900
723
+ },
724
+ {
725
+ "epoch": 0.02,
726
+ "learning_rate": 4.234693877551021e-06,
727
+ "loss": 0.1425,
728
+ "step": 2925
729
+ },
730
+ {
731
+ "epoch": 0.02,
732
+ "learning_rate": 4.183673469387755e-06,
733
+ "loss": 0.1521,
734
+ "step": 2950
735
+ },
736
+ {
737
+ "epoch": 0.02,
738
+ "learning_rate": 4.13265306122449e-06,
739
+ "loss": 0.1434,
740
+ "step": 2975
741
+ },
742
+ {
743
+ "epoch": 0.02,
744
+ "learning_rate": 4.081632653061225e-06,
745
+ "loss": 0.1477,
746
+ "step": 3000
747
+ },
748
+ {
749
+ "epoch": 0.02,
750
+ "eval_loss": 0.13808605074882507,
751
+ "eval_runtime": 13604.3241,
752
+ "eval_samples_per_second": 3.586,
753
+ "eval_steps_per_second": 0.448,
754
+ "eval_wer": 37.97294043601127,
755
+ "step": 3000
756
+ },
757
+ {
758
+ "epoch": 0.02,
759
+ "learning_rate": 4.03061224489796e-06,
760
+ "loss": 0.1383,
761
+ "step": 3025
762
+ },
763
+ {
764
+ "epoch": 0.02,
765
+ "learning_rate": 3.979591836734694e-06,
766
+ "loss": 0.1568,
767
+ "step": 3050
768
+ },
769
+ {
770
+ "epoch": 0.02,
771
+ "learning_rate": 3.928571428571429e-06,
772
+ "loss": 0.1479,
773
+ "step": 3075
774
+ },
775
+ {
776
+ "epoch": 0.02,
777
+ "learning_rate": 3.877551020408164e-06,
778
+ "loss": 0.1197,
779
+ "step": 3100
780
+ },
781
+ {
782
+ "epoch": 0.02,
783
+ "learning_rate": 3.826530612244898e-06,
784
+ "loss": 0.1464,
785
+ "step": 3125
786
+ },
787
+ {
788
+ "epoch": 0.02,
789
+ "learning_rate": 3.7755102040816327e-06,
790
+ "loss": 0.1329,
791
+ "step": 3150
792
+ },
793
+ {
794
+ "epoch": 0.02,
795
+ "learning_rate": 3.724489795918368e-06,
796
+ "loss": 0.1455,
797
+ "step": 3175
798
+ },
799
+ {
800
+ "epoch": 0.02,
801
+ "learning_rate": 3.6734693877551024e-06,
802
+ "loss": 0.138,
803
+ "step": 3200
804
+ },
805
+ {
806
+ "epoch": 0.02,
807
+ "learning_rate": 3.6224489795918373e-06,
808
+ "loss": 0.143,
809
+ "step": 3225
810
+ },
811
+ {
812
+ "epoch": 0.02,
813
+ "learning_rate": 3.5714285714285718e-06,
814
+ "loss": 0.1366,
815
+ "step": 3250
816
+ },
817
+ {
818
+ "epoch": 0.02,
819
+ "learning_rate": 3.5204081632653062e-06,
820
+ "loss": 0.1338,
821
+ "step": 3275
822
+ },
823
+ {
824
+ "epoch": 0.02,
825
+ "learning_rate": 3.469387755102041e-06,
826
+ "loss": 0.1238,
827
+ "step": 3300
828
+ },
829
+ {
830
+ "epoch": 0.02,
831
+ "learning_rate": 3.4183673469387756e-06,
832
+ "loss": 0.1327,
833
+ "step": 3325
834
+ },
835
+ {
836
+ "epoch": 0.02,
837
+ "learning_rate": 3.3673469387755105e-06,
838
+ "loss": 0.1638,
839
+ "step": 3350
840
+ },
841
+ {
842
+ "epoch": 0.02,
843
+ "learning_rate": 3.316326530612245e-06,
844
+ "loss": 0.1426,
845
+ "step": 3375
846
+ },
847
+ {
848
+ "epoch": 0.02,
849
+ "learning_rate": 3.2653061224489794e-06,
850
+ "loss": 0.1057,
851
+ "step": 3400
852
+ },
853
+ {
854
+ "epoch": 0.02,
855
+ "learning_rate": 3.2142857142857147e-06,
856
+ "loss": 0.1491,
857
+ "step": 3425
858
+ },
859
+ {
860
+ "epoch": 0.02,
861
+ "learning_rate": 3.1632653061224496e-06,
862
+ "loss": 0.133,
863
+ "step": 3450
864
+ },
865
+ {
866
+ "epoch": 0.02,
867
+ "learning_rate": 3.112244897959184e-06,
868
+ "loss": 0.1057,
869
+ "step": 3475
870
+ },
871
+ {
872
+ "epoch": 0.02,
873
+ "learning_rate": 3.0612244897959185e-06,
874
+ "loss": 0.1566,
875
+ "step": 3500
876
+ },
877
+ {
878
+ "epoch": 0.02,
879
+ "learning_rate": 3.0102040816326534e-06,
880
+ "loss": 0.1371,
881
+ "step": 3525
882
+ },
883
+ {
884
+ "epoch": 0.02,
885
+ "learning_rate": 2.959183673469388e-06,
886
+ "loss": 0.165,
887
+ "step": 3550
888
+ },
889
+ {
890
+ "epoch": 0.02,
891
+ "learning_rate": 2.908163265306123e-06,
892
+ "loss": 0.1374,
893
+ "step": 3575
894
+ },
895
+ {
896
+ "epoch": 0.02,
897
+ "learning_rate": 2.8571428571428573e-06,
898
+ "loss": 0.1304,
899
+ "step": 3600
900
+ },
901
+ {
902
+ "epoch": 0.02,
903
+ "learning_rate": 2.8061224489795917e-06,
904
+ "loss": 0.1387,
905
+ "step": 3625
906
+ },
907
+ {
908
+ "epoch": 0.02,
909
+ "learning_rate": 2.7551020408163266e-06,
910
+ "loss": 0.1357,
911
+ "step": 3650
912
+ },
913
+ {
914
+ "epoch": 0.02,
915
+ "learning_rate": 2.7040816326530615e-06,
916
+ "loss": 0.156,
917
+ "step": 3675
918
+ },
919
+ {
920
+ "epoch": 0.02,
921
+ "learning_rate": 2.6530612244897964e-06,
922
+ "loss": 0.1475,
923
+ "step": 3700
924
+ },
925
+ {
926
+ "epoch": 0.02,
927
+ "learning_rate": 2.602040816326531e-06,
928
+ "loss": 0.1166,
929
+ "step": 3725
930
+ },
931
+ {
932
+ "epoch": 0.03,
933
+ "learning_rate": 2.5510204081632657e-06,
934
+ "loss": 0.1081,
935
+ "step": 3750
936
+ },
937
+ {
938
+ "epoch": 0.03,
939
+ "learning_rate": 2.5e-06,
940
+ "loss": 0.1409,
941
+ "step": 3775
942
+ },
943
+ {
944
+ "epoch": 0.03,
945
+ "learning_rate": 2.4489795918367347e-06,
946
+ "loss": 0.1125,
947
+ "step": 3800
948
+ },
949
+ {
950
+ "epoch": 0.03,
951
+ "learning_rate": 2.3979591836734696e-06,
952
+ "loss": 0.1142,
953
+ "step": 3825
954
+ },
955
+ {
956
+ "epoch": 0.03,
957
+ "learning_rate": 2.3469387755102044e-06,
958
+ "loss": 0.1318,
959
+ "step": 3850
960
+ },
961
+ {
962
+ "epoch": 0.03,
963
+ "learning_rate": 2.295918367346939e-06,
964
+ "loss": 0.1478,
965
+ "step": 3875
966
+ },
967
+ {
968
+ "epoch": 0.03,
969
+ "learning_rate": 2.244897959183674e-06,
970
+ "loss": 0.1261,
971
+ "step": 3900
972
+ },
973
+ {
974
+ "epoch": 0.03,
975
+ "learning_rate": 2.1938775510204083e-06,
976
+ "loss": 0.1204,
977
+ "step": 3925
978
+ },
979
+ {
980
+ "epoch": 0.03,
981
+ "learning_rate": 2.1428571428571427e-06,
982
+ "loss": 0.1234,
983
+ "step": 3950
984
+ },
985
+ {
986
+ "epoch": 0.03,
987
+ "learning_rate": 2.0918367346938776e-06,
988
+ "loss": 0.1053,
989
+ "step": 3975
990
+ },
991
+ {
992
+ "epoch": 0.03,
993
+ "learning_rate": 2.0408163265306125e-06,
994
+ "loss": 0.1192,
995
+ "step": 4000
996
+ },
997
+ {
998
+ "epoch": 0.03,
999
+ "eval_loss": 0.12595808506011963,
1000
+ "eval_runtime": 14235.6864,
1001
+ "eval_samples_per_second": 3.427,
1002
+ "eval_steps_per_second": 0.428,
1003
+ "eval_wer": 35.44608574071649,
1004
+ "step": 4000
1005
+ }
1006
+ ],
1007
+ "max_steps": 5000,
1008
+ "num_train_epochs": 1,
1009
+ "total_flos": 1.846946562048e+19,
1010
+ "trial_name": null,
1011
+ "trial_params": null
1012
+ }
checkpoint-4000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe99781f928141fde56544c833b4f8a8a1100a885088c1e45942567c2ed1ebfc
3
+ size 3771
checkpoint-5000/config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-small",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 768,
17
+ "decoder_attention_heads": 12,
18
+ "decoder_ffn_dim": 3072,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 12,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 12,
24
+ "encoder_ffn_dim": 3072,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 50257,
28
+ "forced_decoder_ids": null,
29
+ "init_std": 0.02,
30
+ "is_encoder_decoder": true,
31
+ "mask_feature_length": 10,
32
+ "mask_feature_min_masks": 0,
33
+ "mask_feature_prob": 0.0,
34
+ "mask_time_length": 10,
35
+ "mask_time_min_masks": 2,
36
+ "mask_time_prob": 0.05,
37
+ "max_length": 448,
38
+ "max_source_positions": 1500,
39
+ "max_target_positions": 448,
40
+ "model_type": "whisper",
41
+ "num_hidden_layers": 12,
42
+ "num_mel_bins": 80,
43
+ "pad_token_id": 50257,
44
+ "scale_embedding": false,
45
+ "suppress_tokens": [],
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.28.1",
48
+ "use_cache": true,
49
+ "use_weighted_layer_sum": false,
50
+ "vocab_size": 51865
51
+ }
checkpoint-5000/generation_config.json ADDED
@@ -0,0 +1,221 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "begin_suppress_tokens": [
3
+ 220,
4
+ 50257
5
+ ],
6
+ "bos_token_id": 50257,
7
+ "decoder_start_token_id": 50258,
8
+ "eos_token_id": 50257,
9
+ "forced_decoder_ids": [
10
+ [
11
+ 1,
12
+ null
13
+ ],
14
+ [
15
+ 2,
16
+ 50359
17
+ ]
18
+ ],
19
+ "is_multilingual": true,
20
+ "lang_to_id": {
21
+ "<|af|>": 50327,
22
+ "<|am|>": 50334,
23
+ "<|ar|>": 50272,
24
+ "<|as|>": 50350,
25
+ "<|az|>": 50304,
26
+ "<|ba|>": 50355,
27
+ "<|be|>": 50330,
28
+ "<|bg|>": 50292,
29
+ "<|bn|>": 50302,
30
+ "<|bo|>": 50347,
31
+ "<|br|>": 50309,
32
+ "<|bs|>": 50315,
33
+ "<|ca|>": 50270,
34
+ "<|cs|>": 50283,
35
+ "<|cy|>": 50297,
36
+ "<|da|>": 50285,
37
+ "<|de|>": 50261,
38
+ "<|el|>": 50281,
39
+ "<|en|>": 50259,
40
+ "<|es|>": 50262,
41
+ "<|et|>": 50307,
42
+ "<|eu|>": 50310,
43
+ "<|fa|>": 50300,
44
+ "<|fi|>": 50277,
45
+ "<|fo|>": 50338,
46
+ "<|fr|>": 50265,
47
+ "<|gl|>": 50319,
48
+ "<|gu|>": 50333,
49
+ "<|haw|>": 50352,
50
+ "<|ha|>": 50354,
51
+ "<|he|>": 50279,
52
+ "<|hi|>": 50276,
53
+ "<|hr|>": 50291,
54
+ "<|ht|>": 50339,
55
+ "<|hu|>": 50286,
56
+ "<|hy|>": 50312,
57
+ "<|id|>": 50275,
58
+ "<|is|>": 50311,
59
+ "<|it|>": 50274,
60
+ "<|ja|>": 50266,
61
+ "<|jw|>": 50356,
62
+ "<|ka|>": 50329,
63
+ "<|kk|>": 50316,
64
+ "<|km|>": 50323,
65
+ "<|kn|>": 50306,
66
+ "<|ko|>": 50264,
67
+ "<|la|>": 50294,
68
+ "<|lb|>": 50345,
69
+ "<|ln|>": 50353,
70
+ "<|lo|>": 50336,
71
+ "<|lt|>": 50293,
72
+ "<|lv|>": 50301,
73
+ "<|mg|>": 50349,
74
+ "<|mi|>": 50295,
75
+ "<|mk|>": 50308,
76
+ "<|ml|>": 50296,
77
+ "<|mn|>": 50314,
78
+ "<|mr|>": 50320,
79
+ "<|ms|>": 50282,
80
+ "<|mt|>": 50343,
81
+ "<|my|>": 50346,
82
+ "<|ne|>": 50313,
83
+ "<|nl|>": 50271,
84
+ "<|nn|>": 50342,
85
+ "<|no|>": 50288,
86
+ "<|oc|>": 50328,
87
+ "<|pa|>": 50321,
88
+ "<|pl|>": 50269,
89
+ "<|ps|>": 50340,
90
+ "<|pt|>": 50267,
91
+ "<|ro|>": 50284,
92
+ "<|ru|>": 50263,
93
+ "<|sa|>": 50344,
94
+ "<|sd|>": 50332,
95
+ "<|si|>": 50322,
96
+ "<|sk|>": 50298,
97
+ "<|sl|>": 50305,
98
+ "<|sn|>": 50324,
99
+ "<|so|>": 50326,
100
+ "<|sq|>": 50317,
101
+ "<|sr|>": 50303,
102
+ "<|su|>": 50357,
103
+ "<|sv|>": 50273,
104
+ "<|sw|>": 50318,
105
+ "<|ta|>": 50287,
106
+ "<|te|>": 50299,
107
+ "<|tg|>": 50331,
108
+ "<|th|>": 50289,
109
+ "<|tk|>": 50341,
110
+ "<|tl|>": 50348,
111
+ "<|tr|>": 50268,
112
+ "<|tt|>": 50351,
113
+ "<|uk|>": 50280,
114
+ "<|ur|>": 50290,
115
+ "<|uz|>": 50337,
116
+ "<|vi|>": 50278,
117
+ "<|yi|>": 50335,
118
+ "<|yo|>": 50325,
119
+ "<|zh|>": 50260
120
+ },
121
+ "max_initial_timestamp_index": 1,
122
+ "max_length": 448,
123
+ "no_timestamps_token_id": 50363,
124
+ "pad_token_id": 50257,
125
+ "return_timestamps": false,
126
+ "suppress_tokens": [
127
+ 1,
128
+ 2,
129
+ 7,
130
+ 8,
131
+ 9,
132
+ 10,
133
+ 14,
134
+ 25,
135
+ 26,
136
+ 27,
137
+ 28,
138
+ 29,
139
+ 31,
140
+ 58,
141
+ 59,
142
+ 60,
143
+ 61,
144
+ 62,
145
+ 63,
146
+ 90,
147
+ 91,
148
+ 92,
149
+ 93,
150
+ 359,
151
+ 503,
152
+ 522,
153
+ 542,
154
+ 873,
155
+ 893,
156
+ 902,
157
+ 918,
158
+ 922,
159
+ 931,
160
+ 1350,
161
+ 1853,
162
+ 1982,
163
+ 2460,
164
+ 2627,
165
+ 3246,
166
+ 3253,
167
+ 3268,
168
+ 3536,
169
+ 3846,
170
+ 3961,
171
+ 4183,
172
+ 4667,
173
+ 6585,
174
+ 6647,
175
+ 7273,
176
+ 9061,
177
+ 9383,
178
+ 10428,
179
+ 10929,
180
+ 11938,
181
+ 12033,
182
+ 12331,
183
+ 12562,
184
+ 13793,
185
+ 14157,
186
+ 14635,
187
+ 15265,
188
+ 15618,
189
+ 16553,
190
+ 16604,
191
+ 18362,
192
+ 18956,
193
+ 20075,
194
+ 21675,
195
+ 22520,
196
+ 26130,
197
+ 26161,
198
+ 26435,
199
+ 28279,
200
+ 29464,
201
+ 31650,
202
+ 32302,
203
+ 32470,
204
+ 36865,
205
+ 42863,
206
+ 47425,
207
+ 49870,
208
+ 50254,
209
+ 50258,
210
+ 50358,
211
+ 50359,
212
+ 50360,
213
+ 50361,
214
+ 50362
215
+ ],
216
+ "task_to_id": {
217
+ "transcribe": 50359,
218
+ "translate": 50358
219
+ },
220
+ "transformers_version": "4.28.1"
221
+ }
checkpoint-5000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b7ed395ea16a7d80eacfab2f106e681e78116076bb46d24e8b01a6134a1cb0ca
3
+ size 1934161093
checkpoint-5000/preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 80,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
checkpoint-5000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28147f1ab3553e592b72242e2f1da78c7f0985017d382b1cf24ae00065f9e9eb
3
+ size 967102601
checkpoint-5000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7aa0a6912c29db80a45dd19ec66950684a07e55371153ead212c38ac3c0963bd
3
+ size 14575
checkpoint-5000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:64c69d779c30098c55eeeefef22b9e5b75c82c732144663189aae67506842585
3
+ size 627
checkpoint-5000/trainer_state.json ADDED
@@ -0,0 +1,1261 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 29.913363774990813,
3
+ "best_model_checkpoint": "./whisper-small-ko-custom-1000h\\checkpoint-5000",
4
+ "epoch": 0.03346451422911145,
5
+ "global_step": 5000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 2.5e-06,
13
+ "loss": 3.5773,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 5e-06,
19
+ "loss": 1.4784,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.0,
24
+ "learning_rate": 7.500000000000001e-06,
25
+ "loss": 0.753,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.0,
30
+ "learning_rate": 1e-05,
31
+ "loss": 0.5502,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 0.0,
36
+ "learning_rate": 9.948979591836737e-06,
37
+ "loss": 0.3562,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 0.0,
42
+ "learning_rate": 9.89795918367347e-06,
43
+ "loss": 0.2842,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 0.0,
48
+ "learning_rate": 9.846938775510205e-06,
49
+ "loss": 0.2484,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 0.0,
54
+ "learning_rate": 9.795918367346939e-06,
55
+ "loss": 0.2421,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 0.0,
60
+ "learning_rate": 9.744897959183674e-06,
61
+ "loss": 0.2403,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 0.0,
66
+ "learning_rate": 9.693877551020408e-06,
67
+ "loss": 0.2222,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 0.0,
72
+ "learning_rate": 9.642857142857144e-06,
73
+ "loss": 0.202,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 0.0,
78
+ "learning_rate": 9.591836734693878e-06,
79
+ "loss": 0.2197,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 0.0,
84
+ "learning_rate": 9.540816326530612e-06,
85
+ "loss": 0.2045,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 0.0,
90
+ "learning_rate": 9.489795918367348e-06,
91
+ "loss": 0.2275,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 0.0,
96
+ "learning_rate": 9.438775510204082e-06,
97
+ "loss": 0.2023,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 0.0,
102
+ "learning_rate": 9.387755102040818e-06,
103
+ "loss": 0.2212,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 0.0,
108
+ "learning_rate": 9.336734693877552e-06,
109
+ "loss": 0.211,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 0.0,
114
+ "learning_rate": 9.285714285714288e-06,
115
+ "loss": 0.2041,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 0.0,
120
+ "learning_rate": 9.234693877551022e-06,
121
+ "loss": 0.2215,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 0.0,
126
+ "learning_rate": 9.183673469387756e-06,
127
+ "loss": 0.2012,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 0.0,
132
+ "learning_rate": 9.13265306122449e-06,
133
+ "loss": 0.1932,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 0.0,
138
+ "learning_rate": 9.081632653061225e-06,
139
+ "loss": 0.2087,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 0.0,
144
+ "learning_rate": 9.03061224489796e-06,
145
+ "loss": 0.1908,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 0.0,
150
+ "learning_rate": 8.979591836734695e-06,
151
+ "loss": 0.2019,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 0.0,
156
+ "learning_rate": 8.92857142857143e-06,
157
+ "loss": 0.2097,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 0.0,
162
+ "learning_rate": 8.877551020408163e-06,
163
+ "loss": 0.2105,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 0.0,
168
+ "learning_rate": 8.826530612244899e-06,
169
+ "loss": 0.1846,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 0.0,
174
+ "learning_rate": 8.775510204081633e-06,
175
+ "loss": 0.1748,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 0.0,
180
+ "learning_rate": 8.724489795918369e-06,
181
+ "loss": 0.1761,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 0.01,
186
+ "learning_rate": 8.673469387755103e-06,
187
+ "loss": 0.1982,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 0.01,
192
+ "learning_rate": 8.622448979591837e-06,
193
+ "loss": 0.1926,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 0.01,
198
+ "learning_rate": 8.571428571428571e-06,
199
+ "loss": 0.1732,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 0.01,
204
+ "learning_rate": 8.520408163265307e-06,
205
+ "loss": 0.1841,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 0.01,
210
+ "learning_rate": 8.469387755102042e-06,
211
+ "loss": 0.1699,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 0.01,
216
+ "learning_rate": 8.418367346938776e-06,
217
+ "loss": 0.2131,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 0.01,
222
+ "learning_rate": 8.36734693877551e-06,
223
+ "loss": 0.1828,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 0.01,
228
+ "learning_rate": 8.316326530612246e-06,
229
+ "loss": 0.1796,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 0.01,
234
+ "learning_rate": 8.26530612244898e-06,
235
+ "loss": 0.1909,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 0.01,
240
+ "learning_rate": 8.214285714285714e-06,
241
+ "loss": 0.2066,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 0.01,
246
+ "learning_rate": 8.16326530612245e-06,
247
+ "loss": 0.1913,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 0.01,
252
+ "eval_loss": 0.17891158163547516,
253
+ "eval_runtime": 14046.7097,
254
+ "eval_samples_per_second": 3.473,
255
+ "eval_steps_per_second": 0.434,
256
+ "eval_wer": 33.16084938138103,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 0.01,
261
+ "learning_rate": 8.112244897959184e-06,
262
+ "loss": 0.1426,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 0.01,
267
+ "learning_rate": 8.06122448979592e-06,
268
+ "loss": 0.1976,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 0.01,
273
+ "learning_rate": 8.010204081632654e-06,
274
+ "loss": 0.1677,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 0.01,
279
+ "learning_rate": 7.959183673469388e-06,
280
+ "loss": 0.1633,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 0.01,
285
+ "learning_rate": 7.908163265306124e-06,
286
+ "loss": 0.1994,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 0.01,
291
+ "learning_rate": 7.857142857142858e-06,
292
+ "loss": 0.162,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 0.01,
297
+ "learning_rate": 7.806122448979593e-06,
298
+ "loss": 0.1779,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 0.01,
303
+ "learning_rate": 7.755102040816327e-06,
304
+ "loss": 0.1829,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 0.01,
309
+ "learning_rate": 7.704081632653061e-06,
310
+ "loss": 0.1933,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 0.01,
315
+ "learning_rate": 7.653061224489796e-06,
316
+ "loss": 0.1606,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 0.01,
321
+ "learning_rate": 7.602040816326531e-06,
322
+ "loss": 0.1785,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 0.01,
327
+ "learning_rate": 7.551020408163265e-06,
328
+ "loss": 0.1802,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 0.01,
333
+ "learning_rate": 7.500000000000001e-06,
334
+ "loss": 0.17,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 0.01,
339
+ "learning_rate": 7.448979591836736e-06,
340
+ "loss": 0.1679,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 0.01,
345
+ "learning_rate": 7.39795918367347e-06,
346
+ "loss": 0.157,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 0.01,
351
+ "learning_rate": 7.346938775510205e-06,
352
+ "loss": 0.161,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 0.01,
357
+ "learning_rate": 7.295918367346939e-06,
358
+ "loss": 0.1494,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 0.01,
363
+ "learning_rate": 7.244897959183675e-06,
364
+ "loss": 0.1799,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 0.01,
369
+ "learning_rate": 7.193877551020409e-06,
370
+ "loss": 0.1668,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 0.01,
375
+ "learning_rate": 7.1428571428571436e-06,
376
+ "loss": 0.1798,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 0.01,
381
+ "learning_rate": 7.091836734693878e-06,
382
+ "loss": 0.1476,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 0.01,
387
+ "learning_rate": 7.0408163265306125e-06,
388
+ "loss": 0.1779,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 0.01,
393
+ "learning_rate": 6.989795918367348e-06,
394
+ "loss": 0.1795,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 0.01,
399
+ "learning_rate": 6.938775510204082e-06,
400
+ "loss": 0.1642,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 0.01,
405
+ "learning_rate": 6.887755102040817e-06,
406
+ "loss": 0.1239,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 0.01,
411
+ "learning_rate": 6.836734693877551e-06,
412
+ "loss": 0.2013,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 0.01,
417
+ "learning_rate": 6.785714285714287e-06,
418
+ "loss": 0.1646,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 0.01,
423
+ "learning_rate": 6.734693877551021e-06,
424
+ "loss": 0.1339,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 0.01,
429
+ "learning_rate": 6.683673469387756e-06,
430
+ "loss": 0.1814,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 0.01,
435
+ "learning_rate": 6.63265306122449e-06,
436
+ "loss": 0.1302,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 0.01,
441
+ "learning_rate": 6.581632653061225e-06,
442
+ "loss": 0.1324,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 0.01,
447
+ "learning_rate": 6.530612244897959e-06,
448
+ "loss": 0.1384,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 0.01,
453
+ "learning_rate": 6.4795918367346946e-06,
454
+ "loss": 0.1628,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 0.01,
459
+ "learning_rate": 6.4285714285714295e-06,
460
+ "loss": 0.1712,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 0.01,
465
+ "learning_rate": 6.3775510204081635e-06,
466
+ "loss": 0.153,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 0.01,
471
+ "learning_rate": 6.326530612244899e-06,
472
+ "loss": 0.1588,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 0.01,
477
+ "learning_rate": 6.275510204081633e-06,
478
+ "loss": 0.1579,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 0.01,
483
+ "learning_rate": 6.224489795918368e-06,
484
+ "loss": 0.2014,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 0.01,
489
+ "learning_rate": 6.173469387755102e-06,
490
+ "loss": 0.153,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 0.01,
495
+ "learning_rate": 6.122448979591837e-06,
496
+ "loss": 0.141,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 0.01,
501
+ "eval_loss": 0.153132826089859,
502
+ "eval_runtime": 14050.6821,
503
+ "eval_samples_per_second": 3.472,
504
+ "eval_steps_per_second": 0.434,
505
+ "eval_wer": 33.639439546838894,
506
+ "step": 2000
507
+ },
508
+ {
509
+ "epoch": 0.01,
510
+ "learning_rate": 6.071428571428571e-06,
511
+ "loss": 0.1496,
512
+ "step": 2025
513
+ },
514
+ {
515
+ "epoch": 0.01,
516
+ "learning_rate": 6.020408163265307e-06,
517
+ "loss": 0.1456,
518
+ "step": 2050
519
+ },
520
+ {
521
+ "epoch": 0.01,
522
+ "learning_rate": 5.969387755102042e-06,
523
+ "loss": 0.1668,
524
+ "step": 2075
525
+ },
526
+ {
527
+ "epoch": 0.01,
528
+ "learning_rate": 5.918367346938776e-06,
529
+ "loss": 0.1737,
530
+ "step": 2100
531
+ },
532
+ {
533
+ "epoch": 0.01,
534
+ "learning_rate": 5.867346938775511e-06,
535
+ "loss": 0.1378,
536
+ "step": 2125
537
+ },
538
+ {
539
+ "epoch": 0.01,
540
+ "learning_rate": 5.816326530612246e-06,
541
+ "loss": 0.159,
542
+ "step": 2150
543
+ },
544
+ {
545
+ "epoch": 0.01,
546
+ "learning_rate": 5.7653061224489805e-06,
547
+ "loss": 0.1262,
548
+ "step": 2175
549
+ },
550
+ {
551
+ "epoch": 0.01,
552
+ "learning_rate": 5.7142857142857145e-06,
553
+ "loss": 0.1801,
554
+ "step": 2200
555
+ },
556
+ {
557
+ "epoch": 0.01,
558
+ "learning_rate": 5.663265306122449e-06,
559
+ "loss": 0.1221,
560
+ "step": 2225
561
+ },
562
+ {
563
+ "epoch": 0.02,
564
+ "learning_rate": 5.6122448979591834e-06,
565
+ "loss": 0.1261,
566
+ "step": 2250
567
+ },
568
+ {
569
+ "epoch": 0.02,
570
+ "learning_rate": 5.561224489795919e-06,
571
+ "loss": 0.1584,
572
+ "step": 2275
573
+ },
574
+ {
575
+ "epoch": 0.02,
576
+ "learning_rate": 5.510204081632653e-06,
577
+ "loss": 0.1534,
578
+ "step": 2300
579
+ },
580
+ {
581
+ "epoch": 0.02,
582
+ "learning_rate": 5.459183673469388e-06,
583
+ "loss": 0.1572,
584
+ "step": 2325
585
+ },
586
+ {
587
+ "epoch": 0.02,
588
+ "learning_rate": 5.408163265306123e-06,
589
+ "loss": 0.1091,
590
+ "step": 2350
591
+ },
592
+ {
593
+ "epoch": 0.02,
594
+ "learning_rate": 5.357142857142857e-06,
595
+ "loss": 0.1719,
596
+ "step": 2375
597
+ },
598
+ {
599
+ "epoch": 0.02,
600
+ "learning_rate": 5.306122448979593e-06,
601
+ "loss": 0.1817,
602
+ "step": 2400
603
+ },
604
+ {
605
+ "epoch": 0.02,
606
+ "learning_rate": 5.255102040816327e-06,
607
+ "loss": 0.1407,
608
+ "step": 2425
609
+ },
610
+ {
611
+ "epoch": 0.02,
612
+ "learning_rate": 5.204081632653062e-06,
613
+ "loss": 0.1418,
614
+ "step": 2450
615
+ },
616
+ {
617
+ "epoch": 0.02,
618
+ "learning_rate": 5.153061224489796e-06,
619
+ "loss": 0.1489,
620
+ "step": 2475
621
+ },
622
+ {
623
+ "epoch": 0.02,
624
+ "learning_rate": 5.1020408163265315e-06,
625
+ "loss": 0.1542,
626
+ "step": 2500
627
+ },
628
+ {
629
+ "epoch": 0.02,
630
+ "learning_rate": 5.0510204081632655e-06,
631
+ "loss": 0.1643,
632
+ "step": 2525
633
+ },
634
+ {
635
+ "epoch": 0.02,
636
+ "learning_rate": 5e-06,
637
+ "loss": 0.1275,
638
+ "step": 2550
639
+ },
640
+ {
641
+ "epoch": 0.02,
642
+ "learning_rate": 4.948979591836735e-06,
643
+ "loss": 0.1409,
644
+ "step": 2575
645
+ },
646
+ {
647
+ "epoch": 0.02,
648
+ "learning_rate": 4.897959183673469e-06,
649
+ "loss": 0.1463,
650
+ "step": 2600
651
+ },
652
+ {
653
+ "epoch": 0.02,
654
+ "learning_rate": 4.846938775510204e-06,
655
+ "loss": 0.1329,
656
+ "step": 2625
657
+ },
658
+ {
659
+ "epoch": 0.02,
660
+ "learning_rate": 4.795918367346939e-06,
661
+ "loss": 0.1499,
662
+ "step": 2650
663
+ },
664
+ {
665
+ "epoch": 0.02,
666
+ "learning_rate": 4.744897959183674e-06,
667
+ "loss": 0.121,
668
+ "step": 2675
669
+ },
670
+ {
671
+ "epoch": 0.02,
672
+ "learning_rate": 4.693877551020409e-06,
673
+ "loss": 0.1134,
674
+ "step": 2700
675
+ },
676
+ {
677
+ "epoch": 0.02,
678
+ "learning_rate": 4.642857142857144e-06,
679
+ "loss": 0.1321,
680
+ "step": 2725
681
+ },
682
+ {
683
+ "epoch": 0.02,
684
+ "learning_rate": 4.591836734693878e-06,
685
+ "loss": 0.1424,
686
+ "step": 2750
687
+ },
688
+ {
689
+ "epoch": 0.02,
690
+ "learning_rate": 4.540816326530613e-06,
691
+ "loss": 0.1478,
692
+ "step": 2775
693
+ },
694
+ {
695
+ "epoch": 0.02,
696
+ "learning_rate": 4.489795918367348e-06,
697
+ "loss": 0.1289,
698
+ "step": 2800
699
+ },
700
+ {
701
+ "epoch": 0.02,
702
+ "learning_rate": 4.438775510204082e-06,
703
+ "loss": 0.1426,
704
+ "step": 2825
705
+ },
706
+ {
707
+ "epoch": 0.02,
708
+ "learning_rate": 4.3877551020408165e-06,
709
+ "loss": 0.1541,
710
+ "step": 2850
711
+ },
712
+ {
713
+ "epoch": 0.02,
714
+ "learning_rate": 4.336734693877551e-06,
715
+ "loss": 0.1242,
716
+ "step": 2875
717
+ },
718
+ {
719
+ "epoch": 0.02,
720
+ "learning_rate": 4.2857142857142855e-06,
721
+ "loss": 0.1589,
722
+ "step": 2900
723
+ },
724
+ {
725
+ "epoch": 0.02,
726
+ "learning_rate": 4.234693877551021e-06,
727
+ "loss": 0.1425,
728
+ "step": 2925
729
+ },
730
+ {
731
+ "epoch": 0.02,
732
+ "learning_rate": 4.183673469387755e-06,
733
+ "loss": 0.1521,
734
+ "step": 2950
735
+ },
736
+ {
737
+ "epoch": 0.02,
738
+ "learning_rate": 4.13265306122449e-06,
739
+ "loss": 0.1434,
740
+ "step": 2975
741
+ },
742
+ {
743
+ "epoch": 0.02,
744
+ "learning_rate": 4.081632653061225e-06,
745
+ "loss": 0.1477,
746
+ "step": 3000
747
+ },
748
+ {
749
+ "epoch": 0.02,
750
+ "eval_loss": 0.13808605074882507,
751
+ "eval_runtime": 13604.3241,
752
+ "eval_samples_per_second": 3.586,
753
+ "eval_steps_per_second": 0.448,
754
+ "eval_wer": 37.97294043601127,
755
+ "step": 3000
756
+ },
757
+ {
758
+ "epoch": 0.02,
759
+ "learning_rate": 4.03061224489796e-06,
760
+ "loss": 0.1383,
761
+ "step": 3025
762
+ },
763
+ {
764
+ "epoch": 0.02,
765
+ "learning_rate": 3.979591836734694e-06,
766
+ "loss": 0.1568,
767
+ "step": 3050
768
+ },
769
+ {
770
+ "epoch": 0.02,
771
+ "learning_rate": 3.928571428571429e-06,
772
+ "loss": 0.1479,
773
+ "step": 3075
774
+ },
775
+ {
776
+ "epoch": 0.02,
777
+ "learning_rate": 3.877551020408164e-06,
778
+ "loss": 0.1197,
779
+ "step": 3100
780
+ },
781
+ {
782
+ "epoch": 0.02,
783
+ "learning_rate": 3.826530612244898e-06,
784
+ "loss": 0.1464,
785
+ "step": 3125
786
+ },
787
+ {
788
+ "epoch": 0.02,
789
+ "learning_rate": 3.7755102040816327e-06,
790
+ "loss": 0.1329,
791
+ "step": 3150
792
+ },
793
+ {
794
+ "epoch": 0.02,
795
+ "learning_rate": 3.724489795918368e-06,
796
+ "loss": 0.1455,
797
+ "step": 3175
798
+ },
799
+ {
800
+ "epoch": 0.02,
801
+ "learning_rate": 3.6734693877551024e-06,
802
+ "loss": 0.138,
803
+ "step": 3200
804
+ },
805
+ {
806
+ "epoch": 0.02,
807
+ "learning_rate": 3.6224489795918373e-06,
808
+ "loss": 0.143,
809
+ "step": 3225
810
+ },
811
+ {
812
+ "epoch": 0.02,
813
+ "learning_rate": 3.5714285714285718e-06,
814
+ "loss": 0.1366,
815
+ "step": 3250
816
+ },
817
+ {
818
+ "epoch": 0.02,
819
+ "learning_rate": 3.5204081632653062e-06,
820
+ "loss": 0.1338,
821
+ "step": 3275
822
+ },
823
+ {
824
+ "epoch": 0.02,
825
+ "learning_rate": 3.469387755102041e-06,
826
+ "loss": 0.1238,
827
+ "step": 3300
828
+ },
829
+ {
830
+ "epoch": 0.02,
831
+ "learning_rate": 3.4183673469387756e-06,
832
+ "loss": 0.1327,
833
+ "step": 3325
834
+ },
835
+ {
836
+ "epoch": 0.02,
837
+ "learning_rate": 3.3673469387755105e-06,
838
+ "loss": 0.1638,
839
+ "step": 3350
840
+ },
841
+ {
842
+ "epoch": 0.02,
843
+ "learning_rate": 3.316326530612245e-06,
844
+ "loss": 0.1426,
845
+ "step": 3375
846
+ },
847
+ {
848
+ "epoch": 0.02,
849
+ "learning_rate": 3.2653061224489794e-06,
850
+ "loss": 0.1057,
851
+ "step": 3400
852
+ },
853
+ {
854
+ "epoch": 0.02,
855
+ "learning_rate": 3.2142857142857147e-06,
856
+ "loss": 0.1491,
857
+ "step": 3425
858
+ },
859
+ {
860
+ "epoch": 0.02,
861
+ "learning_rate": 3.1632653061224496e-06,
862
+ "loss": 0.133,
863
+ "step": 3450
864
+ },
865
+ {
866
+ "epoch": 0.02,
867
+ "learning_rate": 3.112244897959184e-06,
868
+ "loss": 0.1057,
869
+ "step": 3475
870
+ },
871
+ {
872
+ "epoch": 0.02,
873
+ "learning_rate": 3.0612244897959185e-06,
874
+ "loss": 0.1566,
875
+ "step": 3500
876
+ },
877
+ {
878
+ "epoch": 0.02,
879
+ "learning_rate": 3.0102040816326534e-06,
880
+ "loss": 0.1371,
881
+ "step": 3525
882
+ },
883
+ {
884
+ "epoch": 0.02,
885
+ "learning_rate": 2.959183673469388e-06,
886
+ "loss": 0.165,
887
+ "step": 3550
888
+ },
889
+ {
890
+ "epoch": 0.02,
891
+ "learning_rate": 2.908163265306123e-06,
892
+ "loss": 0.1374,
893
+ "step": 3575
894
+ },
895
+ {
896
+ "epoch": 0.02,
897
+ "learning_rate": 2.8571428571428573e-06,
898
+ "loss": 0.1304,
899
+ "step": 3600
900
+ },
901
+ {
902
+ "epoch": 0.02,
903
+ "learning_rate": 2.8061224489795917e-06,
904
+ "loss": 0.1387,
905
+ "step": 3625
906
+ },
907
+ {
908
+ "epoch": 0.02,
909
+ "learning_rate": 2.7551020408163266e-06,
910
+ "loss": 0.1357,
911
+ "step": 3650
912
+ },
913
+ {
914
+ "epoch": 0.02,
915
+ "learning_rate": 2.7040816326530615e-06,
916
+ "loss": 0.156,
917
+ "step": 3675
918
+ },
919
+ {
920
+ "epoch": 0.02,
921
+ "learning_rate": 2.6530612244897964e-06,
922
+ "loss": 0.1475,
923
+ "step": 3700
924
+ },
925
+ {
926
+ "epoch": 0.02,
927
+ "learning_rate": 2.602040816326531e-06,
928
+ "loss": 0.1166,
929
+ "step": 3725
930
+ },
931
+ {
932
+ "epoch": 0.03,
933
+ "learning_rate": 2.5510204081632657e-06,
934
+ "loss": 0.1081,
935
+ "step": 3750
936
+ },
937
+ {
938
+ "epoch": 0.03,
939
+ "learning_rate": 2.5e-06,
940
+ "loss": 0.1409,
941
+ "step": 3775
942
+ },
943
+ {
944
+ "epoch": 0.03,
945
+ "learning_rate": 2.4489795918367347e-06,
946
+ "loss": 0.1125,
947
+ "step": 3800
948
+ },
949
+ {
950
+ "epoch": 0.03,
951
+ "learning_rate": 2.3979591836734696e-06,
952
+ "loss": 0.1142,
953
+ "step": 3825
954
+ },
955
+ {
956
+ "epoch": 0.03,
957
+ "learning_rate": 2.3469387755102044e-06,
958
+ "loss": 0.1318,
959
+ "step": 3850
960
+ },
961
+ {
962
+ "epoch": 0.03,
963
+ "learning_rate": 2.295918367346939e-06,
964
+ "loss": 0.1478,
965
+ "step": 3875
966
+ },
967
+ {
968
+ "epoch": 0.03,
969
+ "learning_rate": 2.244897959183674e-06,
970
+ "loss": 0.1261,
971
+ "step": 3900
972
+ },
973
+ {
974
+ "epoch": 0.03,
975
+ "learning_rate": 2.1938775510204083e-06,
976
+ "loss": 0.1204,
977
+ "step": 3925
978
+ },
979
+ {
980
+ "epoch": 0.03,
981
+ "learning_rate": 2.1428571428571427e-06,
982
+ "loss": 0.1234,
983
+ "step": 3950
984
+ },
985
+ {
986
+ "epoch": 0.03,
987
+ "learning_rate": 2.0918367346938776e-06,
988
+ "loss": 0.1053,
989
+ "step": 3975
990
+ },
991
+ {
992
+ "epoch": 0.03,
993
+ "learning_rate": 2.0408163265306125e-06,
994
+ "loss": 0.1192,
995
+ "step": 4000
996
+ },
997
+ {
998
+ "epoch": 0.03,
999
+ "eval_loss": 0.12595808506011963,
1000
+ "eval_runtime": 14235.6864,
1001
+ "eval_samples_per_second": 3.427,
1002
+ "eval_steps_per_second": 0.428,
1003
+ "eval_wer": 35.44608574071649,
1004
+ "step": 4000
1005
+ },
1006
+ {
1007
+ "epoch": 0.03,
1008
+ "learning_rate": 1.989795918367347e-06,
1009
+ "loss": 0.155,
1010
+ "step": 4025
1011
+ },
1012
+ {
1013
+ "epoch": 0.03,
1014
+ "learning_rate": 1.938775510204082e-06,
1015
+ "loss": 0.1311,
1016
+ "step": 4050
1017
+ },
1018
+ {
1019
+ "epoch": 0.03,
1020
+ "learning_rate": 1.8877551020408163e-06,
1021
+ "loss": 0.1212,
1022
+ "step": 4075
1023
+ },
1024
+ {
1025
+ "epoch": 0.03,
1026
+ "learning_rate": 1.8367346938775512e-06,
1027
+ "loss": 0.1334,
1028
+ "step": 4100
1029
+ },
1030
+ {
1031
+ "epoch": 0.03,
1032
+ "learning_rate": 1.7857142857142859e-06,
1033
+ "loss": 0.1562,
1034
+ "step": 4125
1035
+ },
1036
+ {
1037
+ "epoch": 0.03,
1038
+ "learning_rate": 1.7346938775510206e-06,
1039
+ "loss": 0.1427,
1040
+ "step": 4150
1041
+ },
1042
+ {
1043
+ "epoch": 0.03,
1044
+ "learning_rate": 1.6836734693877552e-06,
1045
+ "loss": 0.137,
1046
+ "step": 4175
1047
+ },
1048
+ {
1049
+ "epoch": 0.03,
1050
+ "learning_rate": 1.6326530612244897e-06,
1051
+ "loss": 0.1073,
1052
+ "step": 4200
1053
+ },
1054
+ {
1055
+ "epoch": 0.03,
1056
+ "learning_rate": 1.5816326530612248e-06,
1057
+ "loss": 0.1378,
1058
+ "step": 4225
1059
+ },
1060
+ {
1061
+ "epoch": 0.03,
1062
+ "learning_rate": 1.5306122448979593e-06,
1063
+ "loss": 0.1219,
1064
+ "step": 4250
1065
+ },
1066
+ {
1067
+ "epoch": 0.03,
1068
+ "learning_rate": 1.479591836734694e-06,
1069
+ "loss": 0.127,
1070
+ "step": 4275
1071
+ },
1072
+ {
1073
+ "epoch": 0.03,
1074
+ "learning_rate": 1.4285714285714286e-06,
1075
+ "loss": 0.1288,
1076
+ "step": 4300
1077
+ },
1078
+ {
1079
+ "epoch": 0.03,
1080
+ "learning_rate": 1.3775510204081633e-06,
1081
+ "loss": 0.1124,
1082
+ "step": 4325
1083
+ },
1084
+ {
1085
+ "epoch": 0.03,
1086
+ "learning_rate": 1.3265306122448982e-06,
1087
+ "loss": 0.1216,
1088
+ "step": 4350
1089
+ },
1090
+ {
1091
+ "epoch": 0.03,
1092
+ "learning_rate": 1.2755102040816329e-06,
1093
+ "loss": 0.1297,
1094
+ "step": 4375
1095
+ },
1096
+ {
1097
+ "epoch": 0.03,
1098
+ "learning_rate": 1.2244897959183673e-06,
1099
+ "loss": 0.1174,
1100
+ "step": 4400
1101
+ },
1102
+ {
1103
+ "epoch": 0.03,
1104
+ "learning_rate": 1.1734693877551022e-06,
1105
+ "loss": 0.1254,
1106
+ "step": 4425
1107
+ },
1108
+ {
1109
+ "epoch": 0.03,
1110
+ "learning_rate": 1.122448979591837e-06,
1111
+ "loss": 0.1028,
1112
+ "step": 4450
1113
+ },
1114
+ {
1115
+ "epoch": 0.03,
1116
+ "learning_rate": 1.0714285714285714e-06,
1117
+ "loss": 0.1223,
1118
+ "step": 4475
1119
+ },
1120
+ {
1121
+ "epoch": 0.03,
1122
+ "learning_rate": 1.0204081632653063e-06,
1123
+ "loss": 0.136,
1124
+ "step": 4500
1125
+ },
1126
+ {
1127
+ "epoch": 0.03,
1128
+ "learning_rate": 9.69387755102041e-07,
1129
+ "loss": 0.1499,
1130
+ "step": 4525
1131
+ },
1132
+ {
1133
+ "epoch": 0.03,
1134
+ "learning_rate": 9.183673469387756e-07,
1135
+ "loss": 0.1203,
1136
+ "step": 4550
1137
+ },
1138
+ {
1139
+ "epoch": 0.03,
1140
+ "learning_rate": 8.673469387755103e-07,
1141
+ "loss": 0.1053,
1142
+ "step": 4575
1143
+ },
1144
+ {
1145
+ "epoch": 0.03,
1146
+ "learning_rate": 8.163265306122449e-07,
1147
+ "loss": 0.1405,
1148
+ "step": 4600
1149
+ },
1150
+ {
1151
+ "epoch": 0.03,
1152
+ "learning_rate": 7.653061224489796e-07,
1153
+ "loss": 0.1046,
1154
+ "step": 4625
1155
+ },
1156
+ {
1157
+ "epoch": 0.03,
1158
+ "learning_rate": 7.142857142857143e-07,
1159
+ "loss": 0.1136,
1160
+ "step": 4650
1161
+ },
1162
+ {
1163
+ "epoch": 0.03,
1164
+ "learning_rate": 6.632653061224491e-07,
1165
+ "loss": 0.1108,
1166
+ "step": 4675
1167
+ },
1168
+ {
1169
+ "epoch": 0.03,
1170
+ "learning_rate": 6.122448979591837e-07,
1171
+ "loss": 0.1241,
1172
+ "step": 4700
1173
+ },
1174
+ {
1175
+ "epoch": 0.03,
1176
+ "learning_rate": 5.612244897959184e-07,
1177
+ "loss": 0.0961,
1178
+ "step": 4725
1179
+ },
1180
+ {
1181
+ "epoch": 0.03,
1182
+ "learning_rate": 5.102040816326531e-07,
1183
+ "loss": 0.1283,
1184
+ "step": 4750
1185
+ },
1186
+ {
1187
+ "epoch": 0.03,
1188
+ "learning_rate": 4.591836734693878e-07,
1189
+ "loss": 0.1045,
1190
+ "step": 4775
1191
+ },
1192
+ {
1193
+ "epoch": 0.03,
1194
+ "learning_rate": 4.0816326530612243e-07,
1195
+ "loss": 0.1374,
1196
+ "step": 4800
1197
+ },
1198
+ {
1199
+ "epoch": 0.03,
1200
+ "learning_rate": 3.5714285714285716e-07,
1201
+ "loss": 0.111,
1202
+ "step": 4825
1203
+ },
1204
+ {
1205
+ "epoch": 0.03,
1206
+ "learning_rate": 3.0612244897959183e-07,
1207
+ "loss": 0.1293,
1208
+ "step": 4850
1209
+ },
1210
+ {
1211
+ "epoch": 0.03,
1212
+ "learning_rate": 2.5510204081632656e-07,
1213
+ "loss": 0.1048,
1214
+ "step": 4875
1215
+ },
1216
+ {
1217
+ "epoch": 0.03,
1218
+ "learning_rate": 2.0408163265306121e-07,
1219
+ "loss": 0.1322,
1220
+ "step": 4900
1221
+ },
1222
+ {
1223
+ "epoch": 0.03,
1224
+ "learning_rate": 1.5306122448979592e-07,
1225
+ "loss": 0.1229,
1226
+ "step": 4925
1227
+ },
1228
+ {
1229
+ "epoch": 0.03,
1230
+ "learning_rate": 1.0204081632653061e-07,
1231
+ "loss": 0.1301,
1232
+ "step": 4950
1233
+ },
1234
+ {
1235
+ "epoch": 0.03,
1236
+ "learning_rate": 5.1020408163265303e-08,
1237
+ "loss": 0.1258,
1238
+ "step": 4975
1239
+ },
1240
+ {
1241
+ "epoch": 0.03,
1242
+ "learning_rate": 0.0,
1243
+ "loss": 0.0962,
1244
+ "step": 5000
1245
+ },
1246
+ {
1247
+ "epoch": 0.03,
1248
+ "eval_loss": 0.12108764797449112,
1249
+ "eval_runtime": 13144.6817,
1250
+ "eval_samples_per_second": 3.712,
1251
+ "eval_steps_per_second": 0.464,
1252
+ "eval_wer": 29.913363774990813,
1253
+ "step": 5000
1254
+ }
1255
+ ],
1256
+ "max_steps": 5000,
1257
+ "num_train_epochs": 1,
1258
+ "total_flos": 2.30868320256e+19,
1259
+ "trial_name": null,
1260
+ "trial_params": null
1261
+ }
checkpoint-5000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe99781f928141fde56544c833b4f8a8a1100a885088c1e45942567c2ed1ebfc
3
+ size 3771
generation_config.json ADDED
@@ -0,0 +1,221 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "begin_suppress_tokens": [
3
+ 220,
4
+ 50257
5
+ ],
6
+ "bos_token_id": 50257,
7
+ "decoder_start_token_id": 50258,
8
+ "eos_token_id": 50257,
9
+ "forced_decoder_ids": [
10
+ [
11
+ 1,
12
+ null
13
+ ],
14
+ [
15
+ 2,
16
+ 50359
17
+ ]
18
+ ],
19
+ "is_multilingual": true,
20
+ "lang_to_id": {
21
+ "<|af|>": 50327,
22
+ "<|am|>": 50334,
23
+ "<|ar|>": 50272,
24
+ "<|as|>": 50350,
25
+ "<|az|>": 50304,
26
+ "<|ba|>": 50355,
27
+ "<|be|>": 50330,
28
+ "<|bg|>": 50292,
29
+ "<|bn|>": 50302,
30
+ "<|bo|>": 50347,
31
+ "<|br|>": 50309,
32
+ "<|bs|>": 50315,
33
+ "<|ca|>": 50270,
34
+ "<|cs|>": 50283,
35
+ "<|cy|>": 50297,
36
+ "<|da|>": 50285,
37
+ "<|de|>": 50261,
38
+ "<|el|>": 50281,
39
+ "<|en|>": 50259,
40
+ "<|es|>": 50262,
41
+ "<|et|>": 50307,
42
+ "<|eu|>": 50310,
43
+ "<|fa|>": 50300,
44
+ "<|fi|>": 50277,
45
+ "<|fo|>": 50338,
46
+ "<|fr|>": 50265,
47
+ "<|gl|>": 50319,
48
+ "<|gu|>": 50333,
49
+ "<|haw|>": 50352,
50
+ "<|ha|>": 50354,
51
+ "<|he|>": 50279,
52
+ "<|hi|>": 50276,
53
+ "<|hr|>": 50291,
54
+ "<|ht|>": 50339,
55
+ "<|hu|>": 50286,
56
+ "<|hy|>": 50312,
57
+ "<|id|>": 50275,
58
+ "<|is|>": 50311,
59
+ "<|it|>": 50274,
60
+ "<|ja|>": 50266,
61
+ "<|jw|>": 50356,
62
+ "<|ka|>": 50329,
63
+ "<|kk|>": 50316,
64
+ "<|km|>": 50323,
65
+ "<|kn|>": 50306,
66
+ "<|ko|>": 50264,
67
+ "<|la|>": 50294,
68
+ "<|lb|>": 50345,
69
+ "<|ln|>": 50353,
70
+ "<|lo|>": 50336,
71
+ "<|lt|>": 50293,
72
+ "<|lv|>": 50301,
73
+ "<|mg|>": 50349,
74
+ "<|mi|>": 50295,
75
+ "<|mk|>": 50308,
76
+ "<|ml|>": 50296,
77
+ "<|mn|>": 50314,
78
+ "<|mr|>": 50320,
79
+ "<|ms|>": 50282,
80
+ "<|mt|>": 50343,
81
+ "<|my|>": 50346,
82
+ "<|ne|>": 50313,
83
+ "<|nl|>": 50271,
84
+ "<|nn|>": 50342,
85
+ "<|no|>": 50288,
86
+ "<|oc|>": 50328,
87
+ "<|pa|>": 50321,
88
+ "<|pl|>": 50269,
89
+ "<|ps|>": 50340,
90
+ "<|pt|>": 50267,
91
+ "<|ro|>": 50284,
92
+ "<|ru|>": 50263,
93
+ "<|sa|>": 50344,
94
+ "<|sd|>": 50332,
95
+ "<|si|>": 50322,
96
+ "<|sk|>": 50298,
97
+ "<|sl|>": 50305,
98
+ "<|sn|>": 50324,
99
+ "<|so|>": 50326,
100
+ "<|sq|>": 50317,
101
+ "<|sr|>": 50303,
102
+ "<|su|>": 50357,
103
+ "<|sv|>": 50273,
104
+ "<|sw|>": 50318,
105
+ "<|ta|>": 50287,
106
+ "<|te|>": 50299,
107
+ "<|tg|>": 50331,
108
+ "<|th|>": 50289,
109
+ "<|tk|>": 50341,
110
+ "<|tl|>": 50348,
111
+ "<|tr|>": 50268,
112
+ "<|tt|>": 50351,
113
+ "<|uk|>": 50280,
114
+ "<|ur|>": 50290,
115
+ "<|uz|>": 50337,
116
+ "<|vi|>": 50278,
117
+ "<|yi|>": 50335,
118
+ "<|yo|>": 50325,
119
+ "<|zh|>": 50260
120
+ },
121
+ "max_initial_timestamp_index": 1,
122
+ "max_length": 448,
123
+ "no_timestamps_token_id": 50363,
124
+ "pad_token_id": 50257,
125
+ "return_timestamps": false,
126
+ "suppress_tokens": [
127
+ 1,
128
+ 2,
129
+ 7,
130
+ 8,
131
+ 9,
132
+ 10,
133
+ 14,
134
+ 25,
135
+ 26,
136
+ 27,
137
+ 28,
138
+ 29,
139
+ 31,
140
+ 58,
141
+ 59,
142
+ 60,
143
+ 61,
144
+ 62,
145
+ 63,
146
+ 90,
147
+ 91,
148
+ 92,
149
+ 93,
150
+ 359,
151
+ 503,
152
+ 522,
153
+ 542,
154
+ 873,
155
+ 893,
156
+ 902,
157
+ 918,
158
+ 922,
159
+ 931,
160
+ 1350,
161
+ 1853,
162
+ 1982,
163
+ 2460,
164
+ 2627,
165
+ 3246,
166
+ 3253,
167
+ 3268,
168
+ 3536,
169
+ 3846,
170
+ 3961,
171
+ 4183,
172
+ 4667,
173
+ 6585,
174
+ 6647,
175
+ 7273,
176
+ 9061,
177
+ 9383,
178
+ 10428,
179
+ 10929,
180
+ 11938,
181
+ 12033,
182
+ 12331,
183
+ 12562,
184
+ 13793,
185
+ 14157,
186
+ 14635,
187
+ 15265,
188
+ 15618,
189
+ 16553,
190
+ 16604,
191
+ 18362,
192
+ 18956,
193
+ 20075,
194
+ 21675,
195
+ 22520,
196
+ 26130,
197
+ 26161,
198
+ 26435,
199
+ 28279,
200
+ 29464,
201
+ 31650,
202
+ 32302,
203
+ 32470,
204
+ 36865,
205
+ 42863,
206
+ 47425,
207
+ 49870,
208
+ 50254,
209
+ 50258,
210
+ 50358,
211
+ 50359,
212
+ 50360,
213
+ 50361,
214
+ 50362
215
+ ],
216
+ "task_to_id": {
217
+ "transcribe": 50359,
218
+ "translate": 50358
219
+ },
220
+ "transformers_version": "4.28.1"
221
+ }
runs/Apr27_17-49-52_AI-2nd/events.out.tfevents.1682585399.AI-2nd.11600.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:26da90811c0e8b5fa450b1d7741614d0c36063b18902177e5d2400263cb7ec0b
3
- size 37810
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8d45d222ff17493ec490ba4387b6ce5b759a8e9546451351eb45566079f6f9a
3
+ size 38164